Abstract
We consider the following class of quasilinear Schrödinger equations introduced in plasma physics and nonlinear optics with Stein–Weiss convolution parts
where \(\kappa \in \mathbb {R}\backslash \{0\}\) is a parameter, \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\) and H is the primitive of h that fulfills the critical exponential growth in the Trudinger–Moser sense. For \(\kappa <0\): (i) via using a change of variable argument and the mountain-pass theorem, we investigate the existence of ground state solutions only assuming that \(V\in C^0(\mathbb {R}^2,\mathbb {R}^+)\) and \(\inf _{x \in \mathbb {R}^2}V(x)>0\), which complements and generalizes the problems proposed in our recent work in Alves and Shen (J Differ Equ 344:352–404, 2023); (ii) by developing a new type of Trudinger–Moser inequality, we establish a Pohoz̆aev type ground solution by the constraint minimization approach when \(V\equiv 1\). Moreover, if \(\kappa >0\) is small, combining the mountain-pass theorem and Nash–Moser iteration procedure, we obtain the existence of nontrivial solutions, where the asymptotical behavior is also considered when \(\kappa \rightarrow 0^+\). It seems that the results presented above are even new for the case \(\kappa =0\).
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction and main results
In this paper, we are interested in the existence of nontrivial solutions for the nonlocal Schrödinger equations introduced in plasma physics and nonlinear optics with Stein–Weiss convolution parts
where \(\kappa \in \mathbb {R}\backslash \{0\}\) is a parameter, \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\) and H is the primitive of h that fulfills the critical exponential growth in the Trudinger–Moser sense.
If \(\kappa =0\) in Eq. (1.1), it belongs to the so-called Schrödinger equations with Stein–Weiss convolution parts
where \(\beta >0\) and \(0<\mu <2\) and \(0<2\beta +\mu <N\). To treat Eq. (1.2) variationally, the Stein–Weiss inequality which can also be known as the weighted Hardy–Littlewood–Sobolev inequality (HLS in short) usually plays an important role. We recall it below
Proposition 1.1
(see e.g. [55]) Suppose that \(r,s> 1\), \(0<\mu <N\), \(\overline{\beta }+\beta \ge 0\) and \(\overline{\beta }+\beta +\mu \le N\), \(\varphi \in L^r(\mathbb {R}^N)\) and \(\psi \in L^s(\mathbb {R}^N)\). There is a sharp constant \(C=C(\overline{\beta },\beta ,\mu ,N,s,r)>0\), independent of \(\varphi \) and \(\psi \), such that
where
If we suppose that \(\varphi (x)=\psi (x)=|u(x)|^p\) in the weighted Hardy–Littlewood–Sobolev inequality (1.3) with \(\overline{\beta }=\beta \) and \(s=r\) as well as \(2\beta +\mu \le N\), then the integral
is well-defined provided
As pointed out in some previous papers, e.g. [9, 21], for \(N\ge 3\), one could regard \(2^*_{\beta ,\mu }=\frac{2N-2\beta -\mu }{N-2}\) and \(2_{*,\beta ,\mu }=\frac{2N-2\beta -\mu }{N}\) the upper and lower critical Sobolev exponents, respectively. In fact, the critical Sobolev exponents are driven by the following two inequalities
and
Very recently, by introducing a nonlocal version of the concentration-compactness principle, Du et al. [21] considered the existence of nontrivial solutions and investigated the regularity, symmetry of positive solutions by the moving plane arguments for the equation
By developing the Pohoz̆aev identity of
they also studied the existence and non-existence nontrivial solutions with p satisfying (1.4). Moreover, Yang et al. [58] investigated the symmetry, regularity and asymptotical properties as well as sufficient conditions for the nonexistence of nontrivial solutions of semilinear elliptic systems
where \(N\ge 3\) and \(p,q>1\). After the aforementioned works, under the assumptions
- \((\overline{V}_0)\):
-
\(V\in C^0(\mathbb {R}^N)\) and \(\inf _{x\in \mathbb {R}^N}V(x)>0\);
- \((\overline{V}_1)\):
-
there exists a constant \(M_0>0\) such that the set \(\{x\in \mathbb {R}^N:V(x)\le M_0\}\) has a finite Lebesgue measure and \(\Omega = V^{-1}(0)\) is a non-empty set,
Zhang–Tang [61] especially explored the existence and concentration of solutions for the equation
when \(\lambda >0\) is sufficiently large. It is widely known that the potential V with assumptions \((\overline{V}_0)-(\overline{V}_1)\) would be denoted by the steep potential well, see e.g. [13]. In a latest paper [9], by considering Eq. (1.2) with \(V(x)\equiv 1\) and \(N=2\), Alves-Shen combined the mountain-pass theorem and Pohoz̆aev identity to establish the existence of nontrivial solutions, mountaion-pass type solutions, least energy solutions and ground state solutions in the radially symmetric space, where they also studied Eq. (1.2) under the conditions \((\overline{V}_0)\) and
- \((\overline{V}_2)\):
-
\(V (x)\rightarrow \infty \) as \(|x|\rightarrow \infty \); or more generally \(|\{x\in \mathbb {R}^2:V(x)\le M\}|<\infty \) for every \(M>0\); or the function \([V(x)]^{-1}\) belongs to \(L^1(\mathbb {R}^2)\).
The potential V satisfying \((\overline{V}_2)\) is known to be coercive, ses e.g. [4, 19, 44]. In summary, to restore the compactness of Eq. (1.2) caused by the whole Euclidean space \(\mathbb {R}^N\) and the nonlinearity involving critical growth, the usual idea is to suppose \((\overline{V}_1)\), or V being radially symmetric, or \((\overline{V}_2)\). Therefore, one may naturally ask the following two Questions:
- (I):
-
Could we investigate the existence results for Eq. (1.2) by assuming that the positive potential V is periodic?
- (II):
-
Could we even establish the very same results in Question (1) without the periodic assumption on V? In other words, whether we can get the existence results for Eq. (1.2) just by assuming
However, as mentioned in [9, Remark 1.5], the variational functional (even for the periodic potential V) corresponding to Eq. (1.2) does never remain translation invariance in general, which brings a lot of difficulties to solve the problem for a large class of potential V. In our opinion, the Questions (I) and (II) above are mathematically interesting and one of the main purposes in this paper is to explore them. Concerning some other interesting results for the Schrödinger equations with Stein–Weiss convolution parts, we refer to [28, 59, 61, 62] and the references therein.
Let \(\beta =0\) in Eq. (1.2) and \(|x|^{-\mu }\) can be reviewed as the classic Riesz potential, then the nonlinear Schrödinger equation
is closely related to the Choquard equation arising from the study of Bose–Einstein condensation and can be exploited to describe the finite-range many-body interactions between particles, where \(*\) denotes a convolution operator. With respect to the relevant physical case in which \(N = 3\), \(\mu =1\) and \(H(u)=u^2\), Eq. (1.5) turns into the Choquard-Pekar equation which was introduced by Pekar [46] to describe a polaron at rest in the quantum field theory. In [31], Choquard exploited this equation to characterization an electron trapped in its own hole as an approximation to the Hartree-Fock theory for a one component plasma. After a while, by means of the variational methods, Lieb [30] and Lions [33] obtained the existence and uniqueness of positive solutions to (1.5). The regularity, radial symmetry and decay property of the ground state solution were considered in [38, 42]. Equation (1.5) and its variants have received many attentions by many mathematicians because of the appearance of the convolution type nonlinearities over the past decades. We should refer the reader to [1, 7, 8, 42, 53] and the references therein, particular by [43], for a very abundant and meaningful review of the Choquard equations. In fact, Moroz et al. in [41] proposed Eq. (1.5) to be a model for self-gravitating particles in the context as it can be regarded as the Schrödinger–Newton equation.
Then, we would introduce some results on Eq. (1.1) with \(\kappa \ne 0\). Solutions of like it are usually used to the search of certain kinds of standing wave solutions to the nonlinear Schrödinger equation
where \(\psi : \mathbb {R}^2 \times \mathbb {R}\rightarrow \mathbb {C}\), \(\kappa \in \mathbb {R}\backslash \{0\}\), \(W: \mathbb {R}^2\rightarrow \mathbb {R}\) is a given potential and \(\eta : \mathbb {R}^+ \rightarrow \mathbb {R}\) and \(\rho : \mathbb {R}^+ \rightarrow \mathbb {R}\) are appropriate functions. Based on several types of nonlinear term \(\rho (s)\), there are a rich researcher topic in some areas of physics, see e.g. [48]. In the present paper, motivated by [26], we would mainly focus on the superfluid film equation in plasma physics, which corresponds to the case \(\rho (s)=s\). In the meanwhile, it is worthy mentioning here that the scillating soliton instabilities during microwave, laser heating of plasma and so on appeared in nonlinear optics, see [24, 49] for instance.
Due to the real physical applications on Eq. (1.6), there are extensive bibliographies in the study of it by variational methods. To deal with the problem variationally, except in dimension one like [48], the principal barrier is to determine a suitable work space in which the associated energy functional is well-defined and of \(C^1\)-class. By introducing a suitable metric space, authors in [37, 51] exploited the constrain manifold of Nehari type and Nehari–Pohoz̆aev type to study (1.6), respectively. On the other hand, the perturbation procedure developed in [35] was utilized to investigate the quasilinear Schrödinger equations. Besides, there exists another way, namely a change of variable, to overcome the difficult mentioned early. To explain it clearly, we shall split into two cases depending on whether the parameter \(\kappa \) is positive or negative.
With respect to the case in which the parameter \(\kappa <0\), Liu et al. [36] performed the change of variable: \(v=f^{-1}(u)\), where f(t) is defined by
By this change of variable in (1.7), they transformed the quasilinear equation \(-\Delta u+V(x)+\frac{\kappa }{2}u\Delta (u^2)=h(u)\) into the semilinear one \(-\Delta v=f^\prime (v)[h(f(v))-V(x)f(v)]\). Subsequently, with this change of variable to deal with the case \(\kappa <0\), see e.g. [11, 12, 17, 20, 54] ant the references therein.
Obviously, the transformation in (1.7) is unapplicable when \(\kappa >0\). So, it seems that the studies in this direction on the existence results are not as fruitful as the case \(\kappa <0\). In [10], by introducing a new type change of variable, with the help of the mountain-pass theorem and Nash–Moser iteration technique, the authors established the existence of nontrivial solutions when the parameter \(\kappa >0\) is sufficiently small. Along this line, there exist some similar results on quasiliear Schrödinger equations by using this argument, see e.g. [25, 45, 52]. It should be noted that, for \(N\ge 3\), the Sobolev critical exponents are \(22^*=4N/(N-2)\) and \(2^*\) when \(\kappa <0\) and \(\kappa >0\), respectively, see [45] for more details.
Now, we shall turn to the main topics in this paper. The other aim in this work is to consider the existence results for Eq. (1.1) with critical exponential growth by exploiting the variational methods. Motivated by [45] as well as the Trudinger–Moser type inequality, for the cases \(\kappa <0\) and \(\kappa >0\), one can say that a function h possesses critical exponential growth if there is a constant \(\alpha _{0}>0\) such that
and
respectively. This definitions can be found in some literatures, see e.g. [3, 22].
To the best knowledge of us, there exist very few results on Schrödinger equations with Stein–Weiss convolution parts. As explained in [9], the imbedding \(H_0^1(\Omega )\hookrightarrow L^p(\Omega )\) with \(1\le p<+\infty \) can never guarantee \(H_0^1(\Omega )\hookrightarrow L^\infty (\Omega )\) for bounded domain \(\Omega \subset \mathbb {R}^2\). So, one is led to ask if there is another kind of maximal growth in this situation. Indeed, the authors in [40, 47, 56] particularly established the following sharp maximal exponential integrability for functions in \(H_{0}^{1}(\Omega )\):
where the constant \(C=C(\alpha )>0\) and \(|\Omega |\) stands for the Lebesgue measure of \(\Omega \).
There are a lot of generalizations to the well-known Trudinger–Moser inequality in many directions. For example, Adimurthi and Sandeep proved in [5] that the inequality
holds if and only if \(\frac{\alpha }{4\pi }+\frac{s}{2}\le 1\), where \(\alpha >0\) and \(s\in [0,2)\).
Because of the appearance of the singular weight \(|x|^{-\beta }\) in Eq. (1.1), we prefer to use the following version of the Trudinger–Moser inequality in the whole Euclidian space \(\mathbb {R}^2\) established by Adimurthi and Yang [4].
Proposition 1.2
For all \(\alpha >0\), \(s\in [0,2)\) and \(u\in H^1(\mathbb {R}^2)\), there holds
Moreover, for each \(r \in (0,1)\) and \(M>0\), there exists a universal constant \(C=C(r,M)>0\), independent of u, such that
if and only if \(\frac{\alpha }{4\pi }+\frac{s}{2}\le 1\), where
As to some other generalizations, extensions and applications of the Trudinger–Moser inequalities for bounded and unbounded domains, we refer to [22, 29] and the references therein.
Before stating the existence results briefly in the present paper, we shall introduce some notations and definitions. Throughout this paper, by the assumption (V), one can easily find that
is a Hilbert space equipped with the inner product and norm
Obviously, the imbedding \(E\hookrightarrow H^1(\mathbb {R}^2)\) is continuous, where \(H^1(\mathbb {R}^2)\) admits the usual inner product and norm. Let \(L^q(\mathbb {R}^2)\) (\(1\le q\le \infty \)) be the usual Lebesgue space with standard norm \(|u|_q\). Moreover, for all \(s\in [0,2)\), we need the weighted Lebesgue space \(L^q(\mathbb {R}^2,|x|^{-s}dx)\) with \(q\in [1,+\infty )\) defined by
We denote by C and \(C_i\) (\(i=1, 2,\cdots \)) for various positive constants whose exact value may change from lines to lines but are not essential to the analysis of the problem. Let \((X,\Vert \cdot \Vert _X)\) be a Banach space with dual space \((X^{-1},\Vert \cdot \Vert _{X^{-1}})\), and \(\Phi \) be functional on X. The Cerami sequence at a level \(c\in \mathbb {R}\) (\((C)_c\) sequence in short) corresponding to \(\Phi \) means that \(\Phi (x_n)\rightarrow c\) and \((1+\Vert x_n\Vert _X)\Vert \Phi ^{\prime }(x_n)\Vert _{X^{-1}}\rightarrow 0\) as \(n\rightarrow \infty \), where \(\{x_n\}\subset X\).
Firstly, we study the existence results for Eq. (1.1) in the case \(\kappa <0\) under the condition (V):
- (V):
-
\(V\in C^0(\mathbb {R}^2)\) and \(\inf _{x\in \mathbb {R}^2}V(x)>0\).
Moreover, we suppose that the nonlinearity h satisfies (1.8) and the following hypotheses:
- \((h_1)\):
-
\(h\in C(\mathbb {R})\), \(h(s)\equiv 0\) for all \(s\le 0\) and \(h(s)=o(s^{\frac{2-\mu }{2}})\);
- \((h_2)\):
-
there exist some constants \(s_0>0\), \(M_0>0\) and \(\vartheta \in (0,1]\) such that
$$\begin{aligned} 0<s^{\vartheta }H(s)\le M_0h(s),~\forall s\ge s_0; \end{aligned}$$ - \((h_3)\):
-
\(\displaystyle \liminf _{s\rightarrow +\infty }H(s)/e^{\alpha _0s^4}\triangleq \beta _0>0\),
- \((h_4)\):
-
\(h\in C^1(\mathbb {R})\) and there exists a constant \(\delta \in [\frac{1}{2},1)\) such that \(H(s)h^\prime (s)\ge \delta h^2(s)\) for all \(s\ge 0\).
We would like to highlight that a great many functions h satisfying \((h_1)-(h_4)\), for example,
Because \(\kappa <0\) in Eq. (1.1) is arbitrary, without loss of generality, in the sequel we shall always suppose that \(\kappa \equiv -1\) throughout this paper for simplicity. We obtain the following result.
Theorem 1.3
Let (V) and (1.8) with \((h_1)-(h_4)\) be satisfied. Assume that \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\), then Eq. (1.1) admits at least a nontrivial solution \(u_0\in X\) (see Sect. 2 below). If \(\delta =\frac{1}{2}\) in \((h_4)\), then \(\mathcal {J}_f(v_0) = \inf _{v\in \mathcal {N}_f} \mathcal {J}_f(v)\), where \(v_0=f^{-1}(u_0)\) as well as f given in (1.7) and the Nehari manifold \(\mathcal {N}_f\triangleq \{v\in X\backslash \{0\}:\mathcal {J}^\prime _f(v)[v]=0\}\) with \(\mathcal {J}_f:X\rightarrow \mathbb {R}\) defined by
Remark 1.4
We have to point out that the condition, like \((h_4)\), was originally proposed by Cassani and Tarsi in [16], later generalized by Shen, Rădulescu and Yang in [53]. Whereas, it is not obvious to certify that every \((C)_c\) sequence of \(\mathcal {J}_f\) is uniformly bounded because of the appearance of quasilinear term. In particular, compared with [53, Lemma 3.6], we must introduce some new techniques to handle the case \(\delta =\frac{1}{2}\). Moreover, although it fails to cover the case \(\delta \in (0,\frac{1}{2})\), we do not hesitate to confirm that this interval is optimal when the potential V just belongs to \(C^0(\mathbb {R}^2)\). Finally, we succeed in deriving the affirmative answers to the Questions (I) and (II) presented above.
Remark 1.5
Proceeding as some similar arguments in [53, Remark 1.6], we shall also verify some key observations on h and H as follows. On the one hand, by \((h_4)\), one could conclude that \(h^\prime (s)\ge 0\) for all \(s>0\) and then h is nondecreasing on \(s\in (0,+\infty )\) which implies that
On the other hand, one would conclude that \((H(s)/h(s))^\prime \le 1-\delta \) for each \(s>0\) by \((h_4)\). This combined with the fact that \( 0<H(s)=\int _{0}^{s}h(t)dt\) yields
In particular, by \((h_4)\) and (1.13), we could immediately have that
Let us recall that the imbedding \(E\hookrightarrow H^1(\mathbb {R}^2)\) is continuous, then \(E\hookrightarrow L^q(\mathbb {R}^2,|x|^{-\beta }dx)\) is compact since \(H^1(\mathbb {R}^2)\hookrightarrow L^q(\mathbb {R}^2,|x|^{-\beta }dx)\) is compact for every \(q\ge 2\), see Lemma 2.3 below in detail. Thereby, inspired by [9, Theorem 5.8], we suppose that the nonlinearity h satisfies the conditions
- \((H_1)\):
-
For \(\alpha _0>0\) given by (1.8), there exist constants \(b_1,b_2>0\) such that for all \(s\in \mathbb {R}^+\),
$$\begin{aligned} 0<h(s)\le b_1|s|^{\frac{2 -\mu }{2}}+b_2(e^{\alpha _0 s^4}-1); \end{aligned}$$ - \((H_2)\):
-
\(H(s)\le C_0|s|^{\frac{4 -\mu }{2}}+C_0h(s)\) for all \(s\in \mathbb {R}^+\);
- \((H_3)\):
-
\(\tilde{H}(s)\le \tilde{H}(t)\) for all \(0<s<t\), where \(\tilde{H}(s)=h(s)s-2H(s)\);
- \((H_4)\):
-
\(\displaystyle \lim _{|u|\rightarrow \infty }H(s)/|s|^2=+\infty \) in \(x\in \mathbb {R}^2\);
- \((H_5)\):
-
\(\displaystyle \liminf _{s\rightarrow +\infty }sh( s)H( s)e^{-2\alpha _0s^4} \ge \overline{\beta }_0>\displaystyle \inf _{\rho >0}\frac{e^{\frac{4-2\beta -\mu }{4}}V_\rho \rho ^2}{\rho ^{4-2\beta -\mu }\alpha _0^2} \frac{(4-\mu )^2}{(2-\mu )(3-\mu )}\), where \(V_\rho \triangleq \displaystyle \sup _{|x|\le \rho }V(x)\).
As a consequence, we can prove the following result whose proof is omitted.
Corollary 1.6
Let (V) and \((H_1)-(H_5)\) be satisfied. Assume that \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\), then Eq. (1.1) has at least a nontrivial solution in X.
Motivated by [37, 51], we investigate the existence of Pohožaev type ground state solutions for Eq. (1.1) with \(V\equiv 1\) and \(\kappa =-1\), that is,
To the end, instead of \((h_2)-(h_3)\), we need to suppose that h satisfies the following condition
- \((h_5)\):
-
there exist two constants \(p>1\) and a sufficiently large \(\xi >0\) (whose lower bounded can be determined later, see e.g. Theorem 1.8 below) such that \(H(s)\ge \xi s^{p}\) for all \(s\in [0,+\infty )\).
Thus, the second main result in this paper can be stated as follows.
Theorem 1.7
Suppose that h satisfies (1.8), \((h_1)\) and \((h_5)\). If \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\), then Eq. (1.15) admits a ground state solution \(u\in \mathcal {X}\) satisfying
where the corresponding variational \(I:\mathcal {X}\rightarrow \mathbb {R}\) is defined by
For the work space \(\mathcal {X}\) dealt with Eq. (1.15), we would equip it with the following distance
Obviously, the distance is definitely different from those in [37, 51]. Moreover, \((\mathcal {X},d_{\mathcal {X}}(\cdot ,\cdot ))\) is a metric space rather than a vector space because it is not closed under the sum, but it is a complete metric space. Besides, it could be said that \(u\in \mathcal {X}\) is a (weak) solution of Eq. (1.15) if there holds
In a certain sense, the weak solutions of Eq. (1.15) are critical points of I. Combining the assumptions (1.8) and \((h_1)\) as well as Proposition 3.1 below, it can be verified that I is well-defined and of class \(C^0(\mathcal {X})\). Moreover, since \(u+\varphi \in \mathcal {X}\) for every \(u\in \mathcal {X}\) and \(\varphi \in C_{0}^\infty (\mathbb {R}^2)\), the Gateaux derivative of J could be computed as follows
By Lemma 3.2 below, \(u\in \mathcal {X}\) is a weak solution of (1.15) if and only if the Gateaux derivative of I in every direction \(\varphi \in C_{0}^\infty (\mathbb {R}^2)\) vanishes. In view of Lemma 3.3, each weak solution \(u\in \mathcal {X}\backslash \{0\}\) of (1.15) satisfies \(u\in \mathcal {M}\), where
At this position, we shall present some explanations on Theorem 1.7 as follows:
-
(1)
Compared with [9, Theorem 1.4], the main contributions are threefold: (I) the quasilinear term which invokes the well-known Trudinger–Moser inequalities (1.10) and (1.11) in Proposition 1.2 to be unapplicable in Eq. (1.15) is involved; (II) the work space \(\mathcal {X}\) does not need to be radially symmetric; (III) the constraints on h can be greatly relaxed.
-
(2)
In view of the condition \((h_1)\), it behaves as at the original point in the Choquard type setting, see e.g. [53]. As a consequence, to some extent, we shall also regard Eq. (1.15), or (1.1), as the Schrodinger equation of Choquard type with a singular nonlinearity. It is worthy mentioning here that the parameter \(\beta \in (0,2)\) makes sure the imbedding \(H^1(\mathbb {R}^2)\hookrightarrow L^q(\mathbb {R}^2,|x|^{-\beta }dx)\) is compact for every \(q\ge 2\), see Sect. 2 below in detail. Hence, we cannot simply take it for granted that Theorem 1.3, Corollary 1.6 and Theorem 1.7 are also true for \(\beta =0\).
-
(3)
In fact, combining the arguments adopted in [51] and Theorem 1.7, one can also treat Eq. (1.15) with a nonconstant potential after some slight modifications;
-
(4)
To recover the compactness, we fail to take the energy estimate by the assumptions \((h_2)-(h_3)\) since the Moser sequence functions \(\{\overline{w}_n\}\subset \mathcal {X}\) for each fixed \(n\in \mathbb {N}\), but \(\int _{\mathbb {R}^2}|\overline{w}_n|^2|\nabla \overline{w}_n|^2dx\rightarrow +\infty \) as \(n\rightarrow \infty \). Therefore, it would be interesting to construct a new type of Moser sequence functions to overcome this difficulty and we shall contemplate it in a further work.
Finally, we will concentrate on Eq. (1.1) for the case \(\kappa >0\). As stated before, the maximal growth with respect to the nonlinearity h behaves like \(e^{\alpha _0s^2}\) at infinity. So, this is one of essential differences from the case \(\kappa <0\). To establish the existence results, we suppose that
- \((h_6)\):
-
there exist two constants \(\overline{p}>1\) and \(\overline{\xi }>0\) such that \(H(s)\ge \overline{\xi } s^{\overline{p}}\) for all \(s\in [0,1]\);
- \((h_7)\):
-
there exists a constant \(\overline{\eta }>1\) such that \(h(s)s\ge \overline{\eta } H(s)\) for all \(s\ge 0\).
We then establish the third main result in this paper.
Theorem 1.8
Let (V) and (1.9) with \((h_1)\), \((h_6)-(h_7)\) be satisfied. Assume that \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\), if in addition \(\displaystyle \lim _{s\rightarrow 0}h(s)/s=0\) and the constant \(\overline{\xi }>0\) given by \((h_5)\) meets that \(\overline{\xi }>\overline{\xi }_0\) as well as
where \(\overline{V}\triangleq [1+ \max _{x\in B_1(0)}V(x)]\in (0,+\infty )\), then there exists a constant \(\kappa _0>0\) such that Eq. (1.1) possesses a nontrivial solution \(u_\kappa \in E\cap L^\infty (\mathbb {R}^2)\) for all \(\kappa \in (0,\kappa _0)\).
Remark 1.9
It is simple to observe that \((h_3)\) and \((h_4)\) are replaced with \((h_6)\) and \((h_7)\), respectively. We can never suppose \((h_6)\) to be that “\((h^\prime _5):\liminf _{s\rightarrow +\infty }H(s)e^{-\alpha _0 s^2}>0\)", the biggest reason is that it is difficult to verify that Lemma 4.5 remains true if \((h^\prime _5)\) holds. Obviously, \((1-\delta )^{-1}>2\) in \((h_4)\) and \(\eta >1\) in \((h_6)\), but it does not contradict with the depiction “the interval is optimal” in Remark 1.4. Moreover, we prefer to highlight here that the conditions \((h_1)\) and \((h_6)\) in Theorem 1.8 can be replaced with “\(\lim _{s\rightarrow 0^+}H(s)/s^{\eta }\triangleq \overline{\xi }\)" if \(\overline{\xi }>0\) is suitably large. As a by-product of Theorem 1.8, the Questions-(II) mentioned above is fully accomplished to some extent.
Remark 1.10
To our best knowledge, both the results on \(\kappa <0\) and \(\kappa >0\) in Theorems 1.3, 1.7 and 1.8 are new for the Schrödinger equations with Stein–Weiss convolution parts and the nonlinearity involving critical exponential growth. Besides, we believe that our results may prompt further studies on these type equations.
In consideration of Theorem 1.8, the existence of nontrivial solutions of Eq. (1.1) heavily depends on the parameter \(\kappa >0\) being sufficiently small, we wonder what will happen in Theorem 1.8 when \(\kappa \rightarrow 0^+\). In other words, whether \(\{u_\kappa \}\) converges in the sense of a subsequence and what does \(\{u_\kappa \}\) converges to. For these purposes, we study the asymptotical behavior of \(\{u_\kappa \}\) below.
Theorem 1.11
Under the assumptions in Theorem 1.8, let \(u_\kappa \in H^1(\mathbb {R}^2)\cap L^\infty (\mathbb {R}^2)\) be a nontrivial solution of Eq. (1.1), then, passing to a subsequence if necessary, \(u_\kappa \rightarrow u_0\) in \(H^1(\mathbb {R}^2)\) as \(\kappa \rightarrow 0^+\), where \(u_0\) is a nontrivial solution of
As far as we are concerned, it seems the first time to consider the asymptotical behavior of nontrivial solutions for the quasilinear Schrodinger equations in the case \(\kappa >0\) with critical exponential growth.
The paper is organized as follows. By introducing some useful preliminaries which are important in the whole paper, we show the proof of Theorem 1.3 in Sect. 2. Sections 3 and 4 are devoted to the proofs of Theorems 1.7 and 1.8, 1.11, respectively.
2 Preliminaries and the proof of Theorem 1.3
In this section, we try to investigate the existence of nontrivial solutions of Eq. (1.1). Firstly, we observe that Eq. (1.1) is the Euler-Lagrange equation associated with the variational functional
According to the variational point of view, the starting obstacle with respect to Eq. (1.1) is to look for an appropriate function space where J is well-defined. To get though it, we shall make full use of the change of variables introduced in [36]. More precisely, we consider \(v = f ^{-1}(u)\), where f is defined by (1.7) with \(\kappa =-1\). Motivated by [17, 36], one can prove the following result.
Lemma 2.1
f given by (1.7) with \(\kappa \equiv -1\) is an odd function and enjoys the following properties:
- \((f_1)\):
-
\(0<f^\prime (t)\le 1\) for all \(t\in \mathbb {R}\);
- \((f_2)\):
-
\(|f(t)|\le |t|\) for all \(t\in \mathbb {R}\);
- \((f_3)\):
-
\(|f(t)|\le 2^{1/4}|t|^{1/2}\) for all \(t\in \mathbb {R}\);
- \((f_4)\):
-
\(f^\prime (0)=\lim _{t\rightarrow 0}f(t)/t=1\);
- \((f_5)\):
-
\(\lim _{t\rightarrow +\infty }f(t)/\sqrt{t}=2^{1/4}\);
- \((f_6)\):
-
\(\frac{1}{2}f^2(t)\le f(t)f^\prime (t)t\le f^2(t)\) for all \(t\in \mathbb {R}\);
- \((f_7)\):
-
there is a constant \(C>0\) such that \(|f(t)|\ge C|t|\) for \(|t|\le 1\) and \(|f(t)|\ge C\sqrt{|t|}\) for \(|t|\ge 1\);
- \((f_8)\):
-
there exist positive constants \(C_1\) and \(C_2\) satisfying
$$\begin{aligned} |t|\le C_1 |f(t)|+C_2|f(t)|^2,~\forall t\in \mathbb {R}; \end{aligned}$$ - \((f_9)\):
-
\(f^\prime (t)t\) is increasing and \(f(t)f^\prime (t)t^{-1}\) is decreasing for all \(t\in \mathbb {R}\).
After the change of variables \(v = f ^{-1}(u)\), from J , we obtain the following functional \(\mathcal {J}_f=J\circ f:E\rightarrow \mathbb {R}\) defined by
which is well-defined in E and belongs to \(C^1\) under the assumptions (V), (1.8) and \((h_1)\). Moreover, the critical points of \(\mathcal {J}_f\) are weak solutions of the problem
We observe that, if v is nonnegative, then u is nonnegative by \((h_1)\). Hence, to establish the existence of nontrivial solutions of (2.1), the critical point theorem introduced in [14, 39] will be applied.
Proposition 2.2
Let X be a Banach space and \(\Phi \in C^1(X,\mathbb {R})\) Gateaux differentiable for all \(v\in X\), with G-derivative \(\Phi ^\prime (v)\in X^{-1}\) continuous from the norm topology of X to the weak \(*\) topology of \(X^{-1}\) and \(\Phi (0) = 0\). Let S be a closed subset of X which disconnects (archwise) X. Let \(v_0 = 0\) and \(v_1\in X\) be points belonging to distinct connected components of \(E\backslash X\). Suppose that
and let \(\Gamma =\{\gamma \in C([0,1],X):\gamma (0)~\text {and}~\gamma (1)=v_1\}\). Then
and there is a \((C)_c\) sequence for \(\Phi \).
Before verifying that the functional \(\mathcal {J}_f\) admits a mountain-pass geometry, instead of E introduced in Sect. 1, we need to choose a suitable work space. Motivated by [11, 12, 20], we define the space
endowed with the norm
In particular, if \(V(x)\equiv V_0>0\) for all \(x\in \mathbb {R}^2\), the norm above is equivalent to the usual norm in \(H^1(\mathbb {R}^2)\). By (V) and \((f_2)\), then \(E\hookrightarrow X\hookrightarrow H^1(\mathbb {R}^2)\) is continuous which is crucial in this paper. Moreover, by the results in [20], the space \((X,\Vert \cdot \Vert _X)\) is a reflexive and Banach space and \(C_0^\infty (\mathbb {R}^2)\) is dense in it.
Hereafter, we denote by \(\Upsilon >0\) the best constant of the embedding \(X\hookrightarrow H^1(\mathbb {R}^2)\), that is,
Next, we prove an important embedding in our approach.
Lemma 2.3
If \(q\ge 2\) and \(s\in (0, 2)\), then the embedding \(H^1(\mathbb {R}^2)\hookrightarrow L^q(\mathbb {R}^2,|x|^{-s}dx)\) is compact.
Proof
Combining the Egoroff’s theorem and Lebesgue’s Dominated Convergence theorem, the proof is standard and we refer the interested reader to [60, Theorem 1.2] for details. \(\square \)
Lemma 2.4
Let (V) and (1.8) with \((h_1)-(h_4)\) be satisfied. Assume that \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\), then the functional \(\mathcal {J}_f\) possesses the following properties
- (i):
-
there exist two constants \(\varrho ,\rho >0\) such that \(\mathcal {J}_f(v)\ge \varrho \) for all \(v\in X\) with \(\Vert v\Vert _X=\rho \);
- (ii):
-
there exists a function \(e\in X\) with \(\Vert e\Vert _X\ge \rho \) such that \(\mathcal {J}_f(e)<0\).
Proof
(i) Recalling (1.8) and \((h_1)\), for fixed \(\alpha >\alpha _0\), \(q\ge \frac{4-\mu }{2\nu }\) with \(1/\nu +1/\nu ^\prime =1\) and for all \(\varepsilon >0\)
Let \(\varepsilon =1\) in (2.3) with suitable \(\alpha \) and \(\nu ^\prime \), taking \(\Vert v\Vert _X^2<\frac{\pi (4-\mu -2\beta )}{2\Upsilon ^2\alpha _0}\), then we could apply the HLS inequality and (1.11) together with \((f_2)-(f_3)\) and Lemma 2.3 to derive
for \(\epsilon \approx 0^+\). From this, for all \(v\in X\backslash \{0\}\) with \(\Vert v\Vert _X^2<\min \{{1},\frac{\pi (4-2\beta -\mu )}{2\Upsilon ^2\alpha _0}\}\), we obtain
Fixing \(\Vert v\Vert _X=\rho <\frac{1}{4}\) and \(\xi =\frac{4}{\rho }>1\) jointly with \(f(t)t^{-1}\) is decreasing, one gets
Using the Young’s inequality, we arrive at
from where it follows that
and so,
Hence,
there is a sufficiently small constant \(\rho >0\) such that Point-(i) holds.
(ii) From \((h_3)\), we know that
Setting \(\varphi \in C_{0}^{\infty }(\mathbb {R}^2)\) with \(\varphi (x) \ge 0\) for all \(x \in \mathbb {R}^2\), a standard argument gives
then Point-(ii) occurs with \(e=t_0\varphi \) and \(t_0 \approx +\infty \). The proof is complete. \(\square \)
As a consequence of Proposition 2.2 and Lemma 2.4, we obtain the existence of (C) sequence of \(\mathcal {J}_f\) at the level c, that is, \(\mathcal {J}_f(v_n)\rightarrow c\) and \((1+\Vert v_n\Vert _X)\Vert \mathcal {J}_f^\prime (v_n)\Vert _{X^{-1}}\rightarrow 0\). To restore the compactness of \(\{v_n\}\), we firstly derive the upper estimate for the mountain-pass level c. With this aim in mind, we shall deal with it by \((h_2)-(h_3)\). Motivated by [4, 7, 15, 19, 22, 27, 50], we consider the Moser sequence defined by
Lemma 2.5
Let (V) and (1.8) with \((h_1)-(h_4)\) be satisfied. Assume that \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\), then \(0<\varrho \le c<c^*\triangleq \frac{\pi (4-2\beta -\mu )}{4\alpha _0}\).
Proof
Firstly, we obtain \(c\ge \varrho >0\) by Lemma 2.4-(i). Due to Proposition 2.2, one could find that \(c=\inf _{\gamma \in \Gamma }\max _{t\in (0,1]}\mathcal {J}_f(\gamma (t)) \le \inf _{v\in X\backslash \{0\}}\max _{t>0}\mathcal {J}_f(tv)\). So, we must show that there is a function \(w\in X\backslash \{0\}\) such that \(\max _{t>0}\mathcal {J}_f(tw)<c^*\). It follows from some elementary computations that
where
Set \(w_n\triangleq \overline{w}_n/\sqrt{1+\delta _n}\in H ^1(\mathbb {R}^2)\backslash \{0\}\) and since \(\textrm{supp}{\hspace{.05cm}}w_n\subset B_1(0)\), then \(w_n\in X\backslash \{0\}\) and \(\Vert w_n\Vert _{H ^1(\mathbb {R}^2)}\le 1\). We claim that there exists a \(n\in \mathbb {N}^+\) such that
Otherwise, for all \(n\in \mathbb {N}^+\), there is a \(t_n>0\) corresponding to the maximum point of \(\max _{t>0}\mathcal {J}_f(tw_n)\)
in which of the first formula together with \(\Vert w_n\Vert _X\le 1\) implies that
Since \(H(s)\ge 0\) for all \(s\in \mathbb {R}\), by \((f_2)\), we can infer from (2.7) and \(\Vert w_n\Vert _{H ^1(\mathbb {R}^2)}\le 1\) that
Combining \((h_2)\) and \((h_3)\), for all \(\varepsilon \in (0,\beta _0)\), there is a constant \(R_\varepsilon >0\) such that
which together with \((f_5)-(f_6)\) and (2.8)–(2.9) shows that
By (2.9) and \((\vartheta +1)\log (\log n)/2>0\), we can deduce that
If \(\{t_n\}\) is unbounded, up to a subsequence if necessary, we can assume that \(t_n\rightarrow +\infty \) and then
which together with \(\delta _n\rightarrow 0\) in (2.5) yields a contradiction if we tend \(n\rightarrow \infty \). Thereby, passing to a subsequence if necessary, there exists a positive constant \(t_0\) such that
where (2.9) gives the inequality. Moreover, we conclude that \(t_0^2= (4-2\beta -\mu )\pi /2\alpha _0\). Otherwise, we obtain a contradiction by letting \(n\rightarrow \infty \) in (2.11). Let’s tend \(n\rightarrow \infty \) in (2.10), there holds
a contradiction. So, (2.6) holds true. The proof is complete. \(\square \)
Next, we mainly try to conclude that each \((C)_c\) sequence of \(\mathcal {J}_f\) is uniformly bounded. To this aim, we shall split it two cases. In other words, we prefer to divide to \(\delta \in (\frac{1}{2},1)\) and \(\delta =\frac{1}{2}\), respectively. Let us firstly handle the case \(\delta \in (\frac{1}{2},1)\) as follows.
Lemma 2.6
Let (V) and (1.8) with \((h_1)-(h_4)\) be satisfied. Assume that \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\). If \(\delta \in (\frac{1}{2},1)\) in \((h_4)\), then every \((C)_c\) sequence \(\{v_n\}\subset X\) of \(\mathcal {J}_f\) is uniformly bounded.
Proof
Let \(\{v_n\}\subset X\) be a \((C)_c\) sequence of \(\mathcal {J}_f\), that is, \(\mathcal {J}_f(v_n)\rightarrow c\) and \((1+\Vert v_n\Vert _X)\Vert \mathcal {J}^\prime _f(v_n)\Vert _{X^{-1}}\rightarrow 0\), then one immediately obtains
and for all \(\{\psi _n\}\subset X\), there holds
where \(o_n(1)\rightarrow 0\) as \(n\rightarrow \infty \). Without loss of generality, we shall suppose that \(v_n\ne 0\). Furthermore, we can suppose \(v_n>0\) by \((h_1)\). Motivated by [53], we let \(\psi _n= {H(f(v_n))}/[{h(f(v_n))f^\prime (v_n)}]\). Since \(\{v_n\}\subset X\), by using (1.12) and \((f_6)\), one has
and the computation
which together with \((h_4)\), (1.7) and (1.13) gives that
Therefore, \(\{\psi _n\}\subset X\) and it could be applied in (2.13). Moreover, by means of \((h_4)\), (1.7) and (1.13) again,
which jointly with (2.12) and (2.13) indicates that
where we have exploted the fact that \(\Vert \psi _n\Vert _X\le 2\Vert v_n\Vert _X\). By \(\delta \in (\frac{1}{2},1)\), we can conclude that \(\{\mathcal {Q}(v_n)\}\) is bounded. Since
we can accomplish the proof of this lemma. \(\square \)
Conversely, the case \(\delta =\frac{1}{2}\) is totally distinct, and so it is necessary to take a more delicate analysis. In light of this, we need to derive the following result.
Lemma 2.7
Let (V) and (1.8) with \((h_1)-(h_4)\) be satisfied. Assume that \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\). If \(\{v_n\}\subset E\) satisfies \(v_n\rightharpoonup 0\) in E, \(v_n\rightarrow 0\) a.e. in \(\mathbb {R}^2\) and \(|\nabla v_n|^2_2<\frac{\pi (4-2\beta -\mu )}{2\alpha _0}\), then
Proof
Since \(|\nabla v_n|_{2}^2<\frac{\pi (4-2\beta -\mu )}{2\alpha _0}\) and \( X \hookrightarrow H^1(\mathbb {R}^2)\), it follows that \(\Vert v_n\Vert _{H^1(\mathbb {R}^2)}\) is bounded. Thereby, we can argue as (2.4) to obtain
yielding the desired result by Lemma 2.3 for \(\epsilon \approx 0^+\). The proof is complete. \(\square \)
Lemma 2.8
Let (V) and (1.8) with \((h_1)-(h_4)\) be satisfied. Assume that \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\). If \(\delta =\frac{1}{2}\) in \((h_4)\), then every \((C)_c\) sequence \(\{v_n\}\subset X\) of \(\mathcal {J}_f\) is uniformly bounded whenever \(c<c^*\triangleq \frac{\pi (4-2\beta -\mu )}{4\alpha _0}\).
Proof
We argue it by the contradiction and assume, up to a subsequence if necessary, that \(\Vert v_n\Vert _X\rightarrow \infty \), then \(v_n\in E\) for all fixed \(n\in \mathbb {N}^+\) and \(\Vert v_n\Vert \rightarrow \infty \). Define \(\bar{v}_n=\sigma v_n/\Vert v_n\Vert \) with \(\sigma = \sqrt{c+c^* }\), since \(c<c^*\), we have that
Going to a subsequence if necessary, there exists a function \(\bar{v}\in E\) such that \(\bar{v}_n\rightharpoonup \bar{v}\) in E. We claim that \(\bar{v}\ne 0\) in \(\mathbb {R}^2\). Otherwise, we suppose that \(\bar{v}\equiv 0\) a.e. in \(\mathbb {R}^2\). Combining (2.14) and Lemma 2.7,
Define \(\varphi _n \triangleq f( v_n)/f^\prime ( v_n)\) and \(\bar{\varphi }_n \triangleq f( \bar{v}_n)/f^\prime ( \bar{v}_n)\), then \(|\varphi _n|\le 2|v_n|\) and \(|\bar{\varphi }_n|\le 2|\bar{v}_n|\) by \((f_6)\) and
So, \(\Vert \varphi _n \Vert \le 2\Vert v_n\Vert \) and \(\Vert \overline{\varphi }_n \Vert \le 2\Vert \overline{v}_n\Vert \), then \(\varphi _n, \overline{\varphi }_n\in E\). It is simple to compute that \(\max _{t\in (0,1]}\mathcal {J}_f(t\bar{v}_n)\) can be achieved at some \(t_n\in (0,1)\) and then \(\mathcal {J}^\prime _f(t_n\overline{v}_n)[t_n\bar{v}_n]=0\). Going to a subsequence if necessary, we claim that
In fact, we know that \(f( t_n\bar{v}_n)/f^\prime ( t_n\bar{v}_n)\in E\) since \(t_n\in (0,1)\) which makes the left side of (2.16) sense. Setting \(\tilde{v}_n\triangleq t_n \bar{v}_n\), then, selecting a subsequence still denoted by itself, one has \(\tilde{v}_n\rightharpoonup 0\) in E and \(\Vert \tilde{v}_n\Vert ^2\le \Vert \bar{v}_n\Vert ^2=\sigma ^2\). Hence, similar to the proof of Lemma 2.7 and by \((f_6)\) in Lemma 2.1
showing the desired result. Since \(\Vert v_n\Vert \rightarrow +\infty \) as \(n\rightarrow \infty \), \(\sigma t_n/\Vert v_n\Vert \in (0,1)\) for all sufficiently large \(n\in \mathbb {N}\). Combining (1.14), (2.16) and \((f_9)\) in Lemma 2.1, we obtain
Let’s recall that \(\{v_n\}\) is a \((C)_c\) sequence of J, taking the limit \(n\rightarrow \infty \) in (2.17), we would conclude that \(\sigma ^2\le 2c\) by (2.15), a contradiction to (2.14). Consequently, \(\bar{v}\ne 0\) in \(\mathbb {R}^2\) and then there exists a constant \(R>0\) such that \(B_R(0)\cap \mathcal {B}\) admits positive Lebesgue measure, where \(\mathcal {B} \triangleq \{x\in \mathbb {R}^2|\bar{v}(x)\ne 0\}\). Since \(\Vert v_n\Vert \rightarrow \infty \), one knows \(|v_n|\rightarrow \infty \) on \(B_R(0)\cap \mathcal {B}\). It infers from \((h_3)\) that \(H(s)/|s|^2\rightarrow +\infty \) as \(|s|\rightarrow \infty \). Due to \((f_5)\) in Lemma 2.1 and the Fatou’s lemma, we have
Recalling that \(\{v_n\}\) is a \((C)_c\) sequence of \(\mathcal {J}_f\), then
a contradiction. In summary, we finish the proof of this lemma. \(\square \)
Following [9, Lemma 4.6], we have to establish the following result before showing that any weak limit of the \((C)_c\) sequence \(\{v_n\}\subset X\) of \(\mathcal {J}_f\) can be adopted to construct the existence of nontrivial solutions to Eq. (1.1). More precisely, we introduce the following compactness result.
Lemma 2.9
If \(\{v_n\}\subset H^1(\mathbb {R}^2)\) satisfies \(v_n\rightharpoonup v_0\) in \(H^1(\mathbb {R}^2)\) as \(n\rightarrow \infty \) and there exists a constant \(K_0>0\) such that
Then, going to a subsequence if necessary, there holds
Moreover, for all \(\psi \in C_{0}^\infty (\mathbb {R}^2)\), going to a subsequence if necessary, we can conclude that
We must stress here that the ideas in the proof Lemma 2.9 can date back to [9, Lemma 4.6], but there exist some obvious differences because of the quasilinear term and the non-radial symmetrically work space \(H^1(\mathbb {R}^2)\).
Proof
Combining (2.18) and the Fatou lemma, one has
In view of \((h_2)\), one has that
and for all \(\varepsilon >0\), there exists a constant \(\overline{s}=\overline{s}(\varepsilon )>1\) such that
which together with \((f_6)\) in Lemma 2.1, (2.18) and (2.21) gives that
Arguing as in [9, Appendix], we can derive that
where \(\overline{C}>0\) is independent of \(n\in \mathbb {N}\). Let’s define
and similarly
where
Claim 1. \(\{\xi _n\}\) is uniformly bounded in \(n\in \mathbb {N}\) and \(\xi _n\rightarrow \xi _0\) a.e. in \(\mathbb {R}^2\).
Verification: Let \(\varepsilon =1\) and \(q=1\) in (2.3), then one easily sees that there is a constant \(C(\overline{s})>0\) such that
which together with \((f_2)\) in Lemma 2.1 implies that
According to the Holder’s inequality, we can obtain the following two formulas
and
Since
Combining (2.24), (2.25), (2.26) and Lemma 2.3, we deduce that \(\{\xi _n\}\) is uniformly bounded. Thereby, with (2.25) and (2.26) in hand, we can follow [9, Appendix] to show that \(\xi _n\rightarrow \xi _0\) a.e. in \(\mathbb {R}^2\).
Claim 2. \(\Theta _n\rightarrow \Theta _0\) as \(n\rightarrow \infty \).
Verification: Thanks to the HLS inequality and \((f_2)\) in Lemma 2.1,
and \([|x|^{-\mu }*(|x|^{-\beta }|v_n|^{\frac{4-\mu }{2}})] |x|^{-\beta }|v_n|^{\frac{4-\mu }{2}}\rightarrow [|x|^{-\mu }*(|x|^{-\beta }|v_0|^{\frac{4-\mu }{2}})] |x|^{-\beta }|v_0|^{\frac{4-\mu }{2}}\) in \(L^1(\mathbb {R}^2)\) by Lemma 2.3, we apply the generalized Dominated Convergence theorem together with the Claim 1 to obtain
which together with (2.22) yields that
showing the Claim 2 since \(\varepsilon >0\) is arbitrary. In consideration of the given \(\varepsilon >0\), there is a sufficiently large \(n_0\in \mathbb {N}^+\) such that \(|\Theta _n-\Theta _0|\le \varepsilon \) for all \(n\ge n_0\). Now, as a consequence of (2.22)–(2.23), we can conclude that
Thus, (2.19) holds true. Based on the above calculations, the proof of (2.20) is trivial and we could omit it. The proof is complete. \(\square \)
Now, it can be conclude that the variational functional \(\mathcal {J}_f\) admits a nontrivial critical point. In other words, we success in finding a nontrivial solutions for Eq. (1.1). Thus, we can establish the following existence result.
Lemma 2.10
Let (V) and (1.8) with \((h_1)-(h_4)\) be satisfied. Assume that \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\), then Eq. (1.1) has at least a nontrivial solution in X.
Proof
By means of Proposition 2.2, Lemmas 2.4 and 2.5, we know that \(\mathcal {J}_f\) admits a \((C)_c\) sequence, saying it \(\{v_n\}\), in X at the level \(0<c<c^*\). By Lemmas 2.6 and 2.8, one sees that \(\{v_n\}\) is uniformly bounded in X and then (2.18) holds true. Recalling Lemma 2.3, going to a subsequence if necessary, there is a function \(v_0\in X\) such that \(v_n\rightharpoonup v_0\) in X, \(v_n\rightarrow v_0\) in \(L^q(\mathbb {R}^2,|x|^{-s}dx)\) for all \(s\in (0,2)\) and \(q\in [2,+\infty )\), and \(v_n\rightarrow v_0\) a.e. in \(\mathbb {R}^2\). According to (2.19), we have that \(\mathcal {J}_f^\prime (v_0)=0\) which indicates that \(u_0=f(v_0)\) is a solution of Eq. (1.1). By \((f_1)\) in Lemma 2.1, the remaining part is to verify that \(v_0\ne 0\). Arguing it indirectly, we suppose that \(v_0\equiv 0\). By using (2.19) and Lemma 2.5, there holds
Thereby, we shall chose \(\alpha >\alpha _0\) sufficiently close to \(\alpha _0\) and \(\nu ^\prime >1\) sufficiently close to 1 in such a way that \(1/v+1/v^\prime =1\) and
With this choice of \(\alpha >\alpha _0\) and \(\nu >1\), by (1.3), (2.3) and (2.28), we apply (1.11) to derive
for \(\zeta \approx 0^+\). In view of the Cauchy–Schwarz inequality in [32, 9.8 Theorem], or [9, Lemma 5.10], one obtains
which together with \(\mathcal {J}_f^\prime (v_n)[v_n]\rightarrow 0\) yields that \(\Vert v_n\Vert \rightarrow 0\) violating (2.27). The proof is finished. \(\square \)
Nevertheless, we have certified that \(v_0\in E\) is a nontrivial critical point of \(\mathcal {J}_f\), one could never draw the conclusion that \(\mathcal {J}_f(v_0) = \inf _{v\in \mathcal {N}_f} \mathcal {J}_f(v)\). Indeed, without adopting the concentration-compactness principle in the Trudinger–Moser inequality sense developed by Lions [34], later generalized in [18], we would not even conclude that \(\mathcal {J}_f(v_0)\) equals to the mountain-pass value c. For this aim, motivated by [9, Lemma 5.1], we introduce the following energy value associated with \(\mathcal {J}_f\)
Moreover, we can prove that
Lemma 2.11
Let (V) and (1.8) with \((h_1)-(h_4)\) be satisfied. Assume that \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\), for all \(v\in E\backslash \{0\}\), there is a unique \(t_v>0\) such that \(t_vv\in \mathcal {N}_f\) and then \(c_2\triangleq \inf _{v\in \mathcal {N}_f} \mathcal {J}_f(v)=c_1\ge c\).
Proof
Define \(\tau (t)\triangleq \mathcal {J}_f(tv)\) for all \(t>0\), proceeding as the proof of Lemma 2.4, the existence of \(t_v\) is trivial and we omit it here. Next, we concentrate ourself on how to show the uniqueness of \(t_v\). It follows from [57, Theorem 4.1] that
Then, thanks to (1.13) and (1.14) with \(\delta \in [\frac{1}{2},1)\), we can claim that
By \((f_9)\) in Lemma 2.1 and (2.29), it can infer that
is decreasing on \(t\in (0,+\infty )\). Thus, we can deduce that \(t_v\) is unique. To show \(c_2=c_1\), we claim that
To illustrate it, for all \(t>0\) and \(s,s_1,s_2>0\), we define
It follows from some simple calculations that we can get (2.30) if \(\xi (t,s)\ge 0\) and \(\zeta (t,s_1,s_2)\ge 0\) for all \(t>0\) and \(s,s_1,s_2>0\). Firstly, with the help of \((f_9)\) in Lemma 2.1,
which revels that \(\xi (t,s)\ge \min _{t>0}\xi (t,s)=\xi (1,s)=0\). Secondly, we exploit (2.29) to obtain
and so \(\zeta (t,s_1,s_2)\ge \min _{t>0}\zeta (t,s_1,s_2)=\zeta (1,s_1,s_2)=0\). With (2.30) in hand, we immediately have \(c_2\ge c_1\). Alternatively, the existence of \(t_v>0\) is sufficient to show that \(c_2\le c_1\) and \(c\le c_1\). Thereby, we can accomplish the proof of this lemma. \(\square \)
Now, we can present the proof of Theorem 1.3 as follows.
Proof of Theorem 1.3
The remainder is to show \(c=c_2\). To the end, adopting \((f_6)\) and (1.13) with \(\delta =\frac{1}{2}\),
finishing the proof. \(\square \)
3 The proof of Theorem 1.7
In this section, to deal with Eq. (1.15) variationally, what the priority is how to verify that the variational functional I given by (1.16) is well-defined and continuous in \(\mathcal {X}\). Since we consider Eq. (1.15) in \(\mathcal {X}\) instead of \(H^1(\mathbb {R}^2)\), Proposition 1.2 seems to be unapplicable to handle the nonlinearity with a critical exponential growth like (1.8). To begin with, we develop the following version type of Trudinger–Moser inequality.
Proposition 3.1
For all \(\alpha >0\), \(s\in [0,2)\) and \(u\in \mathcal {X}\), there holds
Moreover, there exists a universal constant \(C>0\), independent of u, such that
provided that \(\frac{\alpha }{\pi }+\frac{s}{2}<1\).
Proof
Since \(u\in \mathcal {X}\), then \(u^2\in H^1(\mathbb {R}^2)\). If we replace \(u^2\) with u in (1.10), we obtain (3.1) immediately. To show (3.2), one firstly finds that \(\{u\in \mathcal {X}:d_\mathcal {X}(u,0)\le 1\}\subset \{u \in H^1(\mathbb {R}^2): \Vert u\Vert _{H^1(\mathbb {R}^2)}\le 1\}\). In fact, for every \(u\in \{u\in \mathcal {X}:d_\mathcal {X}(u,0)\le 1\}\), the definition of \(d_\mathcal {X}(u,0)\) reveals that \(4\Vert u\Vert _{H^1(\mathbb {R}^2)}^2+|\nabla u^2|_2^2\le 4\) and so \(\Vert u\Vert _{H^1(\mathbb {R}^2)}^2\le 1\). Moreover, due to the Ladyzhenskaya inequality (see e.g. [23, (II.3.9)]), we obtain
which implies that
Thereby, for all \(u\in \{u\in \mathcal {X}:d_\mathcal {X}(u,0)<1\}\) and \(\frac{\alpha }{\pi }+\frac{s}{2}<1\), then \(\frac{\alpha \Vert u^2\Vert ^2_{H^1(\mathbb {R}^2)}}{4\pi }+\frac{s}{2}<1\). By (1.11),
yielding the desired result. The proof of this proposition is complete. \(\square \)
Lemma 3.2
Suppose that h satisfies (1.8) and \((h_1)\), then the functional I in (1.16) is well-defined and continuous in \(\mathcal {X}\). Moreover, if the Gateaux derivative of I evaluated in \(u\in \mathcal {X}\) is zero in every direction \(\psi \in C_0^\infty (\mathbb {R}^2)\), then u is a weak solution of Eq. (1.15).
Proof
Let \(\varepsilon =1\) in (2.3) with suitable \(\alpha \) and \(\nu ^\prime \), by the HLS inequality and (2.3), for all \(u\in \mathcal {X}\), we exploit (3.1) to get
implying that \(I:\mathbb {R}\rightarrow \mathbb {R}\) is well-defined, where we have used Lemma 2.3 since \(\frac{4\beta }{4-\mu }\in (0,2)\).
Next, we derive that \(I:\mathbb {R}\rightarrow \mathbb {R}\) is continuous on \(\mathcal {X}\). Let us suppose that \(d_\mathcal {X}(u_n,u)\rightarrow 0\) indicating that \(u_n\rightarrow u\) in \(H^1(\mathbb {R}^2)\) and \(|\nabla u_n^2|_2^2\rightarrow |\nabla u^2|_2^2\). Furthermore, one has \(u_n^2\rightarrow u^2\) in \(H^1(\mathbb {R}^2)\). Proceeding as the proof of [44, Proposition 1], there are a subsequence \(\{u_n\}\) still denoted by itself and a function \(v\in \mathcal {X}\) such that \(|u_n(x)|\le v(x)\) a.e. in \(\mathbb {R}^2\). So, we can deduce that there is a constant \(C=C(v)>0\) independent of \(n\in \mathbb {N}\) such that
Repeating the very similar arguments explored in Lemma 2.9, we have that
implying that \(I(u_n)\rightarrow I(u)\). So, \(I\in C^0(\mathcal {X},\mathbb {R})\). The remaining part of this lemma is trivial and we omit it here, the interested reader can refer to [57]. We finish the proof of this lemma. \(\square \)
Lemma 3.3
Suppose that h satisfies (1.8) and \((h_1)\), then every nontrivial nontrivial solution u of Eq. (1.15) belongs to \(\mathcal {M}=\{u\in \mathcal {X}\backslash \{0\}:P(u)=0\}\), where
Proof
Although there is a quasilinear term in Eq. (1.15), we could follow the proof of [9, Theorem 1.6] to obtain that \(u\in W^{2,t}(\mathbb {R}^2)\) for every \(t\in ((1-\epsilon _0)s,s)\) with \(\epsilon _0>0\) being sufficiently small and \(s \in (\frac{2}{\beta +\mu },\frac{2}{\beta })\). With this fact in hand, it is standard to show that \(P(u)\equiv 0\), see e.g. [9, Theorem 1.7] in detail. The proof is complete. \(\square \)
Now, we turn to focus on verifying the necessary properties for the manifold \(\mathcal {M}\).
Lemma 3.4
Suppose that h satisfies (1.8), \((h_1)\) and \((h_5)\). If \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\), then for every \(u\in \mathcal {X}\backslash \{0\}\), there exists a unique \(t_u > 0\) such that \(u_{t_u}=u(t_u^{-1}\cdot )\in \mathcal {M}\). In particular, \(I(u_{t_u})=\max _{t>0}I(u_t)\) and then
Proof
In consideration of \(|\nabla u^2_t|_2=|\nabla u^2|_2\) for every \(t>0\) and Lemma 3.2, it is sufficient to arguing as the proof of [9, Lemma 4.9] to finish the proof of this lemma. \(\square \)
Lemma 3.5
Suppose that h satisfies (1.8), \((h_1)\) and \((h_5)\). If \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\), then there is a \(\xi _0>0\) such that for all \(\xi >\xi _0\), there holds
Proof
Let \(\varphi _0\in \mathcal {X}\) be a cut-off function satisfying \(\varphi _0\in C_{0}^\infty (\mathbb {R}^2)\) defined by \(0\le \varphi _0(x)\le 1\) for every \(x\in \mathbb {R}^2\), \(\varphi _0(x)\equiv 1\) if \(|x|\le 1/2\), \(\varphi _0(x)\equiv 0\) if \(|x|\ge 1\) and \(|\nabla \varphi _0|\le 1\) for all \(x\in \mathbb {R}^2\). So, for all \(\theta ,t>0\),
where \(C_\mu >0\) is a constant dependent of \(\mu \). We choose \(\theta =\min \{\sqrt{\frac{c_*}{2\pi +1}},1\}>0\), and \(|\nabla \varphi _0|_2^2\le \pi \) as well as \(|\varphi _0|\le 1\), by (3.4), we get
It infers from some elementary calculations that
provided that \(\xi >\xi _0\) with \(\xi _0\) defined by
Combining (3.3) and (3.5)–(3.6), we conclude the proof of this lemma. \(\square \)
Lemma 3.6
Suppose that h satisfies (1.8), \((h_1)\) and \((h_5)\). If \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\) then there is a constant \(\varrho >0\) such that \(|u_n|_2\ge \varrho \) for all \(n\in \mathbb {N}\), where \(\{u_n\}\subset \mathcal {M}\) is a minimizing sequence of \(c_\mathcal {M}\). In particular, we have that \(c_{\mathcal {M}}> 0\).
Clearly, \((h_1)\) indicates the condition \(\lim _{s\rightarrow 0}h(s)/s^{2-2\beta -\mu }=0\). We then introduce the following type Trudinger–Moser inequality due to Adachi–Tanaka [2]
in the space \(\mathcal {X}\). It follows from this condition and (1.8) that for all fixed \(\alpha >\alpha _0\) and \(\overline{q}\ge 4-2\beta -\mu \),
Moreover, in consideration of the critical exponential growth (1.8), we must modify some computations used in [9, Lemma 4.10]. For this purpose, we introduce the following sharp Gagliardo-Nirenberg inequality [6]
where \(D^{2,1}(\mathbb {R}^2)=\{v\in L^1(\mathbb {R}^2):|\nabla v|\in L^2(\mathbb {R}^2)\}\).
Now, we can show the proof of Lemma 3.6.
Proof
Suppose, by contradiction, that there is a subsequence, still denoted by itself, \(\{u_n\}\subset \mathcal {M}\) such that \(|u_n|_2\rightarrow 0\) in \(H^1(\mathbb {R}^2)\). According to the definition of \(\mathcal {M}\), then \([|x|^{-\mu }*(|x|^{-\beta }H(u_n))]|x|^{-\beta }H(u_n)\rightarrow 0\) in \(L^1(\mathbb {R}^2)\) which implies that \(\limsup _{n\rightarrow \infty }|\nabla u_n^2|_2^2\le 8c_{\mathcal {M}} <8c_*\). Hence, by Lemma 3.5, we shall choose \(\alpha >\alpha _0\) sufficiently close to \(\alpha _0\) and \(\nu ^\prime >1\) sufficiently close to 1 in such a way that \(1/v+1/v^\prime =1\) and
Let \(v=u_n^2\) in (3.7) and (3.9), respectively. As a consequence of (3.8) and (3.10), we derive
yielding that \(|u_n|_2\ge C>0\), where C is a constant independent of \(n\in \mathbb {N}\), where we have adopted the Ladyzhenskaya inequality in the last second inequality. This is impossible because \(|u_n|_2\rightarrow 0\). The remaining part is trivial and we omit it. The proof of this lemma is finished. \(\square \)
Lemma 3.7
Suppose that h satisfies (1.8), \((h_1)\) and \((h_5)\). If \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\) then the minimization problem \(c_{\mathcal {M}}=\inf _{u\in \mathcal {M}}I(u)\) would be achieved.
Proof
Let \(\{u_n\}\subset \mathcal {M}\) be a minimizing sequence of \(c_\mathcal {M}\), that is, \(I(u_n)\rightarrow c_\mathcal {M}\) and \(P(u_n)=0\). Hence,
revealing that \(\{u_n\}\) is uniformly bounded in \(\mathcal {X}\). In fact, recalling Lemma 3.5, we can further obtain
Going to a subsequence if necessary, there is a function \(u\in \mathcal {X}\) such that \(u_n\rightharpoonup u\) in \(H^1(\mathbb {R}^2)\), \(u_n^2\rightharpoonup u^2\) in \(H^1(\mathbb {R}^2)\), \(u_n \rightarrow u\) in \(L^q(\mathbb {R}^2,|x|^{-s}dx)\) for all \(s\in (0,2)\) with \(q\in [2,+\infty )\), and \(u_n\rightarrow u\) a.e. in \(\mathbb {R}^2\). To conclude that \(d_\mathcal {X}(u_n,u)\rightarrow 0\), we claim that
Indeed, by means of \(d_\mathcal {X}(u_n,0)<1\) for all sufficiently large \(n\in \mathbb {N}\) in (3.11), we have that
For the fixed \(\alpha >\alpha _0\), \(q\ge \frac{4-\mu }{2\nu }\) with \(1/\nu +1/\nu ^\prime =1\) in (2.3), letting \(\varepsilon =1\) with suitable \(\alpha \) and \(\nu ^\prime \), there holds \(\frac{4\alpha \nu ^\prime d^2_\mathcal {X}(u_n,0)}{\pi }+\frac{s}{2}<1\) with \(s=\frac{4\beta }{4-\mu }\in (0,2)\), we can apply (3.2) with denoting \(v_n=\frac{u_n}{\sqrt{d_\mathcal {X}(u_n,0)}}\) (which gives that \(d_\mathcal {X}(v_n,0)\le 1\)) to obtain for each measurable set \(\Omega \subset \mathbb {R}^2\) the inequality below
Therefore, setting the numbers \(l_1=\frac{4-\mu }{4}\), \(l_2=\frac{4-\mu }{4\nu }\) and the sequences
we have that
and
for all measurable set \(\Omega \subset \mathbb {R}^2\). This jointly with the Vitali’s Dominated Convergence theorem to get (3.12). With the help of (3.12) and the Fatou’s lemma,
Combining the first part of Lemma 3.6 and (3.12), we immediately conclude that \(u\ne 0\). Therefore, by means of (3.13), there exists a constant \(t_0\in (0,1]\) such that \(P(u_{t_0})=0\). Using Fatou’s lemma again, we obtain
indicating the desired result. The proof of this lemma is complete. \(\square \)
We then try to certify that every attained function of the minimization problem \(c_{\mathcal {M}}=\inf _{u\in \mathcal {M}}I(u)\). Before proceeding this aim, we have two claims listed as follows.
Claim A. For all \(u\in \mathcal {X}\), there holds \(d_{\mathcal {X}}(u_t,u)\rightarrow 0\) as \(t\rightarrow 1\).
Verification: Since \(C_0^\infty (\mathbb {R}^2)\) is dense in \(H^1(\mathbb {R}^2)\), for each \(\varepsilon >0\), there exist \(U \in C^\infty _0(\mathbb {R}^2,\mathbb {R}^2)\) and \(v \in C^\infty _0(\mathbb {R}^2)\) such that
By some simple computations
and
Recalling that \(|\nabla u_t^2|_2^2=|\nabla u^2|_2^2\) for all \(t>0\), by letting \(t\rightarrow 1\), we deduce that
showing the claim.
Claim B. By Lemma 3.4, the map \(\mathcal {X}\backslash \{0\}\rightarrow (0,+\infty )\) defined by \(u\mapsto t_u\) is continuous.
Verification: To finish it, we firstly have the following assertion:
where \(t_n\triangleq t_{u_n}>0\) satisfies \((u_n)_{t_n}=u_n(t_n^{-1}\cdot )\in \mathcal {M}\) (Lemma 3.4 derives the existence of \(t_n\). For each fixed \(t>0\), thanks to [44, Proposition 1], we can follow the proof of Lemma 3.2 to derive
Given \(\varepsilon \in (0,1)\), define \(s_1=1-\varepsilon \) and \(s_2=1+\varepsilon \). Since \(u_0\in \mathcal {M}\), then Lemma 3.4 reveals that \(t=1\) is the unique maximum point of \(I((u_0)_t)\), and therefore \(\frac{d}{dt}|_{t=s_1}I((u_0)_{t})>0\) and \(\frac{d}{dt}|_{t=s_2}I((u_0)_{t})<0\). So, by using (3.15), for some sufficiently large \(n\in \mathbb {N}\), one has
which indicate that \(1-\varepsilon<t_n<1+\varepsilon \). So, (3.14) holds true.
Let \(\{u_n\}\subset \mathcal {X}\backslash \{0\}\) be a sequence satisfying \(d_\mathcal {X}(u_n,u_0)\) as \(n\rightarrow \infty \), there is a unique \(t_u>0\) such that \(u_{t_u}\in \mathcal {M}\). Denoting \(v_n\triangleq (u_n)_{t_u}\), then \(d_\mathcal {X}(v_n,u_{t_u})\) as \(n\rightarrow \infty \). In view of (3.14), we have \(\overline{t}_n\rightarrow 1\) as \(n\rightarrow \infty \), where \((v_n)_{\overline{t}_n}\in \mathcal {M}\). On the other hand, it is easy to prove that, passing to a subsequence if necessary, \(t_n\rightarrow t_0\) as \(n\rightarrow \infty \), where \((u_n)_{t_n}\in \mathcal {M}\). By the uniqueness in Lemma 3.4, one deduces \(\overline{t}_nt_u=t_n\) which indicates that \(t_0=t_u\) by tending \(n\rightarrow \infty \) on both sides. So, the claim is true.
Lemma 3.8
Suppose that h satisfies (1.8), \((h_1)\) and \((h_5)\). If \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\), then any minimizer of \(c_{\mathcal {M}}=\inf _{u\in \mathcal {M}}I(u)\) is a solution of Eq. (1.15).
Proof
Arguing it indirectly, we could suppose that u is not a weak solution of Eq. (1.15) and there exists a function \(\varphi \in C_{0}^\infty (\mathbb {R}^2)\) such that \(I^\prime (u)[\varphi ]<0\). Fixing \(\epsilon >0\) such that
Due to Claim A, there exists a constant \(\delta =\min \{\delta (u,\epsilon ), \delta (\varphi ,\epsilon )\}>0\) such that
Set \(Q^\vartheta \triangleq u+\vartheta \varphi \in \mathcal {X}\), thereby there exists a unique \(t_{Q^\vartheta }>0\) such that \(Q^\vartheta _{t_{Q^\vartheta }}\in \mathcal {M}\) by Lemma 3.4. As \(u\in \mathcal {M}\), one has that \(t_u=1\) by Lemma 3.4. Obviously, \(d_{\mathcal {X}}(Q^\vartheta ,u) \rightarrow 0\) as \(\vartheta \rightarrow 0\), hence \(t_{Q^\vartheta }\rightarrow 1\) as \(\vartheta \rightarrow 0\) by Claim B. For \(\vartheta \in (0,\epsilon )\) sufficiently small, we can derive that \(|t_{Q^\vartheta }-1|<\delta \) which together with (3.17) gives that
Consequently, by (3.18) we may define \(w\triangleq u_{t_{Q^\vartheta }}-u\) and \(\overline{\varphi }\triangleq \varphi _{t_{Q^\vartheta }}\) in (3.16),
which together with \(Q^\vartheta _{t_{Q^\vartheta }}\in \mathcal {M}\) yields a contradiction. The proof is complete. \(\square \)
We are in a position to show the proof of Theorem 1.7.
Proof of Theorem 1.7
The set \(\mathcal {M}\) is well-defined by Lemma 3.4 and then we could consider the minimization problem \(c_\mathcal {M}=\inf _{u\in \mathcal {M}}I(u)\). Combining Lemmas 3.7 and 3.8, one concludes that the minimizer, say it \(u\in \mathcal {X}\), is indeed a nontrivial solution of Pohoz̆aev type ground state of Eq. (1.15). According to (3.3), we can finish the proof. \(\square \)
4 The proofs of Theorems 1.8 and 1.11
In this section, we are mainly concerned with the case \(\kappa >0\) for Eq. (1.1), where the existence of nontrivial solutions and the asymptotical behavior are investigated. Similar to Sect. 2, we have to determine a suitable work space and then consider the problems variationally. Motivated by [10, 25], we firstly study the auxiliary equation
where the continuous even function \(g_\kappa :\mathbb {R}\rightarrow \mathbb {R}\) is defined by
Observe that if we get a solution u of Eq. (4.1) satisfying \(|u|_\infty <1/\sqrt{3\kappa }\), then u is indeed a solution of the original Eq. (1.1). Consequently, to find a nontrivial solution with the prescribed \(L^\infty \)-estimate we shall make full use of the mountain-pass theorem together with the Mash-Moser iteration method. To our best knowledge, to exploit the Mash-Moser iteration successfully, we must take some more delicate analysis than that of in [25] because of the appearance of the Stein–Weiss convolution type nonlinearity with critical exponential growth.
From now on, throughout this section we rewrite \(g_\kappa \) by g just for simplicity. Proceeding as [10, 25], we can present some properties on g below.
Lemma 4.1
For all \(\kappa >0\), then
- \((g_1)\):
-
\(g\in C^1\) is increasing on \((-\infty ,0)\) and decreasing on \((0,+\infty )\) as well as
$$\begin{aligned} 1/\sqrt{6}\le g(t)\le 1~\text {and}~|g^\prime (t)|\le \sqrt{\kappa /2},~\forall t\in \mathbb {R}; \end{aligned}$$ - \((g_2)\):
-
\(-\frac{1}{2}\le t\frac{g^\prime (t)}{g(t)}\le 0\), for all \(t\in \mathbb {R}\);
- \((g_3)\):
-
The primitive \(G(t)=\int _0^tg(\tau )d\tau \) of g(t) is an increasing function and therefore inevitable;
- \((g_4)\):
-
\(t\le G^{-1}(t)\le \sqrt{6}t\) for all \(t\ge 0\), \(\displaystyle \lim _{t\rightarrow 0}\frac{G^{-1}(t)}{t}=1\) and \(\displaystyle \lim _{t\rightarrow +\infty }\frac{G^{-1}(t)}{t}=\sqrt{6}\).
Generally speaking, the (weak) solutions of Eq. (4.1) are critical points of its corresponding energy functional \(I_\kappa :E\rightarrow \mathbb {R}\)
However, \(I_\kappa \) may be not well-defined in E. With Lemma 4.1, we can get across it.
Let \(u=G^{-1}(v)\), then we introduce the energy functional \(\mathcal {J}_g=I_\kappa \circ G^{-1}:E\rightarrow \mathbb {R}\) defined by
In fact, the critical points of \(\mathcal {J}_g\) are solutions of
Thanks to the discussions in [10, 25], we conclude that \(v\in E\) is a solution of Eq. (4.2) if and only if \(u=G^{-1}(v)\) is a solution of Eq. (4.1).
Next, we would formulate the functional setting for a variational approach to Eq. (4.2). Because the nonlinearity f satisfies (1.9) and \((h_1)\), for fixed \(\alpha >\alpha _0\), \(q>1\) and for every \(\varepsilon >0\), we have
and
Given a function \(v\in E\), by (1.3), we utilized \((g_4)\) and (4.4) with \(\alpha >\alpha _0\) and \(q\ge \frac{4-\mu }{2\nu }\) to obtain
where we have adopted (1.10) in Proposition 1.2 together with \(\nu >1\) and \({1}/{\nu }+{1}/{\nu ^\prime }=1\). Consequently, the energy functional \(\mathcal {J}_g\) is well-defined and J is of class \(\mathcal {C}^1\). Moreover, for all \(\psi \in C_0^\infty (\mathbb {R}^2)\),
To look for the mountain-pass type solutions for Eq. (4.2), firstly, we have to establish the validity of the mountain-pass geometry required in Proposition 2.2 for \(\mathcal {J}_g\).
Lemma 4.2
Let (V) and (1.9) with \((h_1)\), \((h_6)-(h_7)\) be satisfied. Assume that \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\), then the functional \(\mathcal {J}_g\) admits the following properties
- (i):
-
there exist two constants \(\varrho _\kappa ,\rho _\kappa >0\) such that \(J_g(v)\ge \varrho _\kappa \) for all \(v\in E\) with \(\Vert v\Vert =\rho _\kappa \);
- (ii):
-
there exists a function \(v_1\in E\) with \(\Vert v_1\Vert \ge \rho _\kappa \) such that \(\mathcal {J}_g(v_1)<0\).
Proof
(i). Recalling (4.5), given a \(v\in E\) with \(\Vert v\Vert ^2<\frac{\pi (4-2\beta -\mu )}{6\alpha _0}\), we use \((g_4)\) and (1.11) to obtain
Since \(4-\mu >2\) and \(q>1\), we can choose a suitably small \(\rho _\kappa ^2<\frac{\pi (4-2\beta -\mu )}{6\alpha _0}\) to get the Point (i).
(ii). Let \(\psi \in C^\infty _0(\mathbb {R}^2)\) with \(|\psi |_\infty \le 1\), by \((h_5)\) and \((g_4)\),
since \(\overline{p}>1\). Letting \(v_1=t_0\psi \) with \(t_0>0\) large enough, we can finish the proof of this lemma. \(\square \)
Via Proposition 2.2 and Lemma 4.2, there is a (C) sequence \(\{v_n\}\subset E\) for \(\mathcal {J}_g\) at the level \(c_\kappa >0\). Although the mountain-pass value \(c_\kappa \) is dependent of \(\kappa >0\), we can take an upper bound for it.
Lemma 4.3
Let (V) and (1.9) with \((h_1)\), \((h_6)-(h_7)\) be satisfied. Assume that \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\), then there is a \(\overline{\xi }_0>0\) such that for every \(\overline{\xi }>\overline{\xi }_0\), there exists a \(\overline{c}=\overline{c}(\overline{\xi })>0\) satisfying \(c_\kappa \le \overline{c}<\frac{\pi (4-2\beta -\mu )(\overline{\eta }-1)}{12\alpha _0\overline{\eta }}\).
Proof
Let \(\varphi _0\in C_0^\infty (\mathbb {R}^2)\) be a cutoff function used in Lemma 3.5. By \((g_4)\) and \((h_6)\),
In particular, one can deduce from (4.6) that
Choosing \(\gamma _0(t)=t\varphi _0\), one easily deduces that \(\gamma _0\in \Gamma _\kappa =\{\gamma \in C([0,1],E):\gamma (0)=0,\mathcal {J}_g(\gamma (1))<0\}\). According to the definition of \(c_k\), then by (4.7) \((g_4)\) and \((h_6)\), we have
By using the assumption on \(\xi \), there holds
So, we can accomplish the proof of this lemma. \(\square \)
Now, with the help of Lemma 4.3, we can obtain the following result.
Lemma 4.4
Let (V) and (1.9) with \((h_1)\), \((h_6)-(h_7)\) be satisfied. Assume that \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\), then any \((C)_{c_\kappa }\) sequence of \(\mathcal {J}_g\) is uniformly bounded.
Proof
Let \(\{v_n\}\subset E\) be a \((C)_{c_\kappa }\) sequence of \(\mathcal {J}_g\), namely \(\mathcal {J}_g(v_n)\rightarrow c_\kappa \) and \((1+\Vert v_n\Vert )\Vert \mathcal {J}^\prime _g(v_n)\Vert _{E^{-1}}\rightarrow 0\) as \(n\rightarrow \infty \). Inspired by [10, 25], we choose \(\psi _n=G^{-1}(v_n)g(G^{-1}(v_n))\). By (V) and Lemma 4.1,
and
showing that \(\{\psi _n\}\subset E\) and so \(\psi _n\) could be a candidate for the test function.
Combining \((g_4)\) and \((h_7)\), it is simple to see that
which together with Lemma 4.3 implies the desired result. The proof is complete. \(\square \)
As a byproduct of Lemma 4.4, up to a subsequence if necessary, there is a function \(v_\kappa \in E\) such that \(v_n\rightharpoonup v_\kappa \) in E, \(v_n\rightarrow v_\kappa \) in \(L^p(\mathbb {R}^2,|x|^{-s}dx)\) for all \(s\in (0,2)\) with \(p\ge 2\) and \(v_n\rightarrow v_\kappa \) a.e. in \(\mathbb {R}^2\).
Next, we show that the energy functional \(J_g\) satisfies the so-called \((C)_{c_\kappa }\) condition.
Lemma 4.5
Let (V) and (1.9) with \((h_1)\), \((h_6)-(h_7)\) be satisfied. Assume that \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\), then \(\{v_n\}\) contains a strongly convergent subsequence for all fixed \(\kappa >0\).
Proof
Due to \((h_1)\), without loss of generality, we can suppose that \(v_n\ge 0\) in \(\mathbb {R}^2\). Since \(\{v_n\}\subset E\) is a \((C)_{c_\kappa }\) sequence of \(\mathcal {J}_g\), combining Lemma 4.3 and (4.4), we could obtain \(\displaystyle \limsup _{n\rightarrow \infty }\Vert v_n\Vert ^2<\frac{\pi (4-2\beta -\mu )}{6\alpha _0}\). Then, letting \(\varepsilon =1\) in (4.3) and (4.4), we choose \(\alpha >\alpha _0\) sufficiently close to \(\alpha _0\) and \(\nu ^\prime >1\) sufficiently close to 1 in such a way that \(1/v+1/v^\prime =1\) and
for all large \(n\in \mathbb {N}\). Hence, we apply (4.3), and (4.9) to get
where \(\epsilon \approx 0^+\), \(q=\nu \ge \frac{4-\mu }{2}\) and Lemma 2.3 are used. Note that \(1/q+1/\nu ^\prime =1\), by means of the Hölder’s inequality,
Recalling Proposition 1.2 and Lemma 2.3, one admits \(v_n\rightarrow v_\kappa \) in \(L^{2}(\mathbb {R}^2,|x|^{ -\frac{4\beta }{4-\mu } }dx)\) and \(L^{\frac{8q}{4-\mu }}(\mathbb {R}^2,|x|^{ -\frac{4\beta }{4-\mu } }dx)\). It follows from the above two formulas as well as (4.9) that
Similarly, we can derive
Let us recall that \((1+\Vert v_n\Vert )\Vert \mathcal {J}^\prime _g(v_n)\Vert _{E^{-1}}\rightarrow 0\), then invoking (4.10)–(4.11) and \((g_1)-(g_2)\), one could find that
showing that \(v_n\rightarrow v_\kappa \) in E, where \(\vartheta _n=\theta v_n+(1-\theta )v_\kappa \) is generated by the Mean Value theorem for some \(\theta \in (0,1)\). So, the proof of this lemma is finished. \(\square \)
As a consequence, we can conclude that the weak limit \(v_\kappa \in E\) is a nontrivial solution of Eq. (4.2).
Lemma 4.6
Let (V) and (1.9) with \((h_1)\), \((h_6)-(h_7)\) be satisfied. Assume that \(\beta >0\), \(0<\mu <2\) with \(0<2\beta +\mu <2\), then Eq. (4.2) admits a nonnegative solution \(v_\kappa \in E\) satisfying \(\Vert v_\kappa \Vert ^2\le \frac{2\overline{\eta }}{\overline{\eta }-1}c_\kappa \).
Proof
By Proposition 2.2 and Lemma 4.2, there exists a (C) sequence \(\{v_n\}\subset E\) for \(\mathcal {J}_g\) at the level \(c_\kappa >0\) (see Lemma 4.2-(i)). According to Lemmas 4.3, 4.4 and 4.5, we obtain that \(\mathcal {J}_g(v_\kappa )=c_\kappa >0\) and \(\mathcal {J}_g^\prime (v_\kappa )=0\) which reveal that \(v_\kappa \) is a nontrivial solution of Eq. (4.2). Proceeding as (4.8), the estimate on \(\Vert v_\kappa \Vert \) is immediate. Thus, we finish the proof of this lemma. \(\square \)
So far, we could deduce that \(u_\kappa = G^{-1}(v_\kappa )\) is a nontrivial solution of Eq. (4.1). Whereas, we must consider the \(L^\infty \)-estimate for \(u_k\) and so it would perhaps be a nontrivial solution of the original Eq. (1.1). By exploiting \((g_4)\) in Lemma 4.1, it suffices to investigate the \(L^\infty \)-estimate for \(v_k\). Arguing as [10, 25], we reach this aim by the Nash–Moser iteration technique.
Nevertheless, different from those nonlinearities explored in [10, 25], we have to face some additional difficulties caused by the Stein–Weiss convolution parts. Consequently, we need to verify the following result.
Lemma 4.7
Under the assumptions in Lemma 4.6 and let \(v_\kappa \in E\) be a nonnegative solution of Eq. (4.2) established by Lemma 4.6, then there is a constant \(C_0>0\) independent of \(\kappa \) such that
Proof
Recalling Lemma 4.5, thanks to [44, Proposition 1], up to a subsequence if necessary, there is a function \(\varpi \in E\) such that \(|v_n(x)|\le \varpi (x)\) and \(|v_\kappa (x)|\le \varpi (x)\) a.e. in \(\mathbb {R}^2\). So, let \(\varepsilon =1\) in (4.4), we have that
To finish the proof, we show that the above integral is well-defined. In view of the Claim 1 in Lemma 2.9, we can find a constant \(C_1^\varpi >0\) such that
We continue to follow the proof of Claim 1 in Lemma 2.9 to have
and since \(|x-y|^{-\mu }\le 1\) whenever \(|x-y|>1\)
Thereby, there is a constant \(C_2^\varpi >0\) such that
Now, we can accomplish the proof of this lemma by choosing \(C_0=C_1^\varpi +C_2^\varpi \in (0,+\infty )\). \(\square \)
Lemma 4.8
Under the assumptions in Lemma 4.6 and let \(v_\kappa \in E\) be the nonnegative solution of Eq. (4.2) established by Lemma 4.6. If in addition \(\lim _{s\rightarrow 0^+}\frac{h(s)}{s}=0\), then \(v_\kappa \in L^\infty (\mathbb {R}^2)\) and
where \(\overline{C}>0\) is a constant independent of \(\kappa \) and \(\tilde{q}\in (1,2)\) with \(\frac{1}{\tilde{q}}+\frac{1}{\tilde{q}^\prime }=1\).
Proof
Let \(\gamma >1\) and \(m\in \mathbb {N}^+\) and we define the sets \(A_m\triangleq \{x\in \mathbb {R}^2:v_\kappa ^{\gamma -1}\le m\}\) and \(B_m=\mathbb {R}^2\backslash A_m\). Consider the sequences
It’s simple to see that \((v_\kappa )_m,(w_\kappa )_m\in E\), \(|(v_\kappa )_m|\le |v_\kappa |^{2\gamma -1}\) and \(|(w_\kappa )_m|^2=v_\kappa (v_\kappa )_m\le |v_\kappa |^{2\gamma }\) in \(\mathbb {R}^2\). Moreover,
which imply that
Combining (4.12) and the fact that \(\gamma >1\), one obtains
Because \(v_\kappa \in E\) is a nontrivial critical point of \(\mathcal {J}_g\), that is, \(\mathcal {J}^\prime _g(v_\kappa )[(v_\kappa )_m]=0\) which gives that
By Lemma 4.7, we have
Since \(h(s)=o(s)\) as \(s\rightarrow 0^+\), by (1.9), there is a constant \(\overline{p}>1\) and \(C_{\alpha ,\overline{p}}\) dependent of \(\alpha \) and \(\overline{p}\) such that
Note that \(v_\kappa (v_\kappa )_m=(w_\kappa )_m^2\) and \(V(x)\ge V_0\) by (V), by using (4.13), (4.14) and (4.15) with \(\overline{p}=2\) to get
which together with the facts \(1/\sqrt{6}\le g(s)\le 1\) and \(s\le G^{-1}(s)\le \sqrt{6}s\) for all \(s\ge 0\) gives that
We fix \(\tilde{q}\in (1,2)\) with \(\tilde{q}^\prime =\widetilde{q}/(\widetilde{q}-1)\) and \(E\hookrightarrow L^4(\mathbb {R}^2)\), there is a constant \(\overline{C}>0\) independent \(\gamma \) and \(\kappa \) such that
where
Once \((w_\kappa )_m=v_\kappa ^\gamma \) in \(A_m\) and \((w_\kappa )_m\le v_\kappa ^\gamma \) in \(\mathbb {R}^2\), there holds
By applying the Dominated Convergence theorem with \(m\rightarrow \infty \) to the above formula, one has
We choose the constant \(\sigma =2/\tilde{q}\), then \(\sigma >1\) because \(\tilde{q}\in (1,2)\). For every \(j\in \mathbb {N}^+\), define \(\gamma _j=\sigma ^j\) and thus \(2\tilde{q}\gamma _{j+1}=2\tilde{q}\sigma \gamma _j=4\gamma _j\). For \(j=1\), \(\gamma _1=\sigma >1\) which can be applied in (4.16) to derive
For \(j=2\), \(\gamma _2=\sigma ^2>1\) and \(2\tilde{q}\gamma _2=4\gamma _1=4\sigma \) and by (4.16),
For \(j=3\), \(\gamma _3=\sigma ^3>1\) and \(2\tilde{q}\gamma _3=4\gamma _2=4\sigma ^2\) and by (4.16),
Similar to (4.17), (4.18) and (4.19), proceeding this iteration procedure j times, we can infer that
invoking that \(v_\kappa \in L^{4\sigma ^j}(\mathbb {R}^2)\) for all \(j\in \mathbb {N}^+\). Clearly, \(\sum _{i=1}^\infty \frac{i}{\sigma ^i}=\frac{\sigma }{(\sigma -1)^2}\) and \(\sum _{i=1}^\infty \frac{1}{\sigma ^i}=\frac{1}{\sigma -1}\), thereby we can take the limit in (4.20) as \(j\rightarrow \infty \) to obtain
finishing the proof of this lemma. \(\square \)
At this step, we can present the proof of Theorem 1.8.
Proof of Theorem 1.8
Combining Lemmas 4.6 and 4.8, we know that Eq. (4.2) has a nonnegative solution \(v_\kappa \in E\) satisfying
where \(\varpi \in E\) is given by Lemma 4.7. So, we can define \(\kappa _0=1/(18d_0^2)\) and then for all \(\kappa \in (0,\kappa _0)\)
In summary, \(u_\kappa \) is a nontrivial nonnegative solution of Eq. (4.1). By (4.21), we recall the definition of \(g=g_\kappa \) to conclude that \(u_\kappa \in E\cap L^\infty (\mathbb {R}^2)\) is a nontrivial nonnegative solution of Eq. (1.1). \(\square \)
Finally, we study the asymptotical behavior of \(u_\kappa \) as \(\kappa \rightarrow 0^+\).
Proof of Theorem 1.11
Since \(u_\kappa =G^{-1}(v_\kappa )\) is a solution of Eq. (1.1), via Lemmas 4.3 and 4.6,
So, going to a subsequence if necessary, there is a function \(u_0\in E\) such that \(u_\kappa \rightharpoonup u_0\) in E, \(u_\kappa \rightarrow u_0\) in \(L^p(\mathbb {R}^2,|x|^{-s}dx)\) for each \(p\in (2,+\infty )\) with \(s\in (0,1)\), and \(u_\kappa \rightarrow u_0\) a.e. in \(\mathbb {R}^2\) as \(\kappa \rightarrow 0^+\). Recalling that \(u_\kappa \) is a solution of Eq. (1.1), for all \(\psi \in C_0^\infty (\mathbb {R}^2)\), there holds
In view of \(|u_\kappa |_\infty \le \sqrt{6}d_0\) in (4.21) and (4.22), as \(\kappa \rightarrow 0^+\), we obtain
In particular, choosing \(\psi =u_\kappa -u_0\in E\) and so \(|\psi |_\infty \) and \(|\nabla \psi |_2\) are uniformly bounded in \(\kappa \in (0,\kappa _0)\) by (4.21) and (4.22). Inserting \(\psi =u_\kappa -u_0\) into (4.24), as \(\kappa \rightarrow 0^+\), we have that
Next, we claim that,
and
Indeed, we argue as the proof of Claim 1 in Lemma 2.9 to derive that \([|x|^{-\mu }*(|x|^{-\beta }H(u_\kappa )) ]|x|^{-\beta } h(u_\kappa )u_\kappa \) \(\rightarrow [|x|^{-\mu }*(|x|^{-\beta }H(u_0)) ]|x|^{-\beta } h(u_0)u_0\) a.e. in \(\mathbb {R}^2\). Therefore, letting \(\varepsilon =1\) in (4.3), we choose \(\alpha >\alpha _0\) sufficiently close to \(\alpha _0\) and \(\nu ^\prime >1\) sufficiently close to 1 in such a way that \(1/v+1/v^\prime =1\) and
for all sufficiently small \(\kappa \in (0,\kappa _0)\). Using \((h_6)\), the HLS inequality, (4.3) and (1.11) with (4.28),
for \(\epsilon \approx 0^+\), which indicates (4.26) by Lemma 2.3 and the generalized Lebesgue’s Dominated Convergence theorem. One can get (4.27) in a very similar way. We tend \(\kappa \rightarrow 0^+\) in (4.23) as well as with (4.24) and (4.27) to derive
showing that \(u_0\in E\) is a solution of Eq. (1.17). Moreover, with the help of (4.23) for \(\psi =u_\kappa -u_0\), one would conclude that \(u_\kappa \rightarrow u_0\) in E by (4.24) and (4.26)–(4.27). Eventually, to accomplish the proof, we have to verify that \(u_0\ne 0\). To the end, letting \(\psi =u_\kappa \) in (4.23), by (4.29) with \(q=\frac{(4-\mu )^2}{4}\),
yielding that \(\Vert u_\kappa \Vert \ge C>0\) since \(\mu <2\), where \(C>0\) is independent of \(\kappa \in (0,\kappa _0)\). Therefore, we have that \(\Vert u_0\Vert \ge C>0\) by \(u_\kappa \rightarrow u_0\) in E. The proof is finished. \(\square \)
Data availability
Data sharing is not applicable to this article as no datasets were generated or analysed during the current study
References
Ackermann, N.: On a periodic Schrödinger equation with nonlocal superlinear part. Math. Z. 248, 423–443 (2004)
Adachi, S., Tanaka, K.: Trudinger type inequalities in \(\mathbb{R} ^N\) and their best exponents. Proc. Am. Math. Soc. 128, 2051–2057 (2000)
Adimurthi, S.L., Yadava, S.L.: Multiplicity results for semilinear elliptic equations in bounded domain of \(\mathbb{R} ^{2}\) involving critical exponent. Ann. Scuola Norm. Sup. Pisa Cl. Sci. (4) 17(4), 481–504 (1990)
Adimurthi, Y.Y.: An interpolation of Hardy inequality and Trudinger–Moser inequality in \(\mathbb{R} ^N\) and its applications. Int. Math. Res. Not. IMRN 13, 2394–2426 (2010)
Adimurthi, K.S.: A singular Moser–Trudinger embedding and its applications. NoDEA Nonlinear Differ. Equ. Appl. 13(5–6), 585–603 (2007)
Agueh, M.: Sharp Gagliardo–Nirenberg inequalities via \(p\)-Laplacian type equations. NoDEA Nonlinear Differ. Equ. Appl. 15, 457–472 (2008)
Alves, C.O., Cassani, D., Tarsi, C., Yang, M.: Existence and concentration of ground state solutions for a critical nonlocal Schrödinger equation in \(\mathbb{R} ^2\). J. Differ. Equ. 261, 1933–1972 (2016)
Alves, C.O., Nobrega, A.B., Yang, M.: Multi-bump solutions for Choquard equation with deepening potential well. Calc. Var. Partial Differ. Equ. 55(3), 1–28 (2016)
Alves, C.O., Shen, L.: Critical Schrödinger equations with Stein–Weiss convolution parts in \(\mathbb{R} ^2\). J. Differ. Equ. 344, 352–404 (2023)
Alves, C.O., Wang, Y., Shen, Y.: Soliton solutions for a class of quasilinear Schrödinger equations with a parameter. J. Differ. Equ. 259(1), 318–343 (2015)
Alves, C.O., Figueiredo, G.M., Severo, U.B.: Multiplicity of positive solutions for a class of quasilinear problems. Adv. Differ. Equ. 252, 911–942 (2009)
Alves, C.O., Figueiredo, G.M., Severo, U.B.: A result of multiplicity of solutions for a class of quasilinear equations. Proc. Edinb. Math. Soc. 55, 291–309 (2012)
Bartsch, T., Pankov, A., Wang, Z.-Q.: Nonlinear Schrödinger equations with steep potential well. Commun. Contemp. Math. 3, 549–569 (2001)
Brézis, H., Nirenberg, L.: Remarks on finding critical points points. Commun. Pure Appl. Math. 44, 939–963 (1991)
Cao, D.: Nontrivial solution of semilinear elliptic equation with critical exponent in \(\mathbb{R} ^2\). Commun. Partial Differ. Equ. 17, 407–435 (1992)
Cassani, D., Tarsi, C.: Schrödinger–Newton equations in dimension two via a Pohozaev–Trudinger log-weighted inequality. Calc. Var. Partial Differ. Equ. 60(5), 1–31 (2021)
Colin, M., Jeanjean, L.: Solutions for a quasilinear Schrödinger equation: a dual approach. Nonlinear Anal. 56(2), 213–226 (2004)
de Souza, M., do João Marcos, B.: A sharp Trudinger–Moser type inequality in \(\mathbb{R} ^2\). Trans. Am. Math. Soc. 366, 4513–4549 (2014)
do João Marcos, B.: \(N\)-Laplacian equations in \(\mathbb{R} ^N\) with critical growth. Abstr. Appl. Anal. 2, 301–315 (1997)
do João Marcos, B., Severo, U.: Solitary waves for a class of quasilinear Schrödinger equations in dimension two. Calc. Var. Partial Differ. Equ. 38, 275–315 (2010)
Du, L., Gao, F., Yang, M.: On elliptic equations with Stein–Weiss type convolution parts. Math. Z. 301, 2185–2225 (2022)
de Figueiredo, D.G., Miyagaki, O.H., Ruf, B.: Elliptic equations in \(\mathbb{R} ^2\) with nonlinearities in the critical growth range. Calc. Var. Partial Differ. Equ. 3, 139–153 (1995)
Galdi, G.P.: An Introduction to the Mathematical Theory of the Navier–Stokes Equations, Steady-State Problems. Springer Monographs in Mathematics, 2nd edn. Springer, New York (2011)
Goldman, M.V.: Strong turbulence of plasma waves. Rev. Modern Phys. 56, 709–735 (1984)
Gloss, E., Severo, U.: Soliton solutions for a class of Schrödinger equations with a positive quasilinear term and critical growth. Proc. Edinb. Math. Soc. (2) 65(1), 279–301 (2022)
Kurihura, S.: Large-amplitude quasi-solitons in superfluid films. J. Phys. Soc. Jpn. 50, 3262–3267 (1981)
Lam, N., Lu, G.: Existence and multiplicity of solutions to equations of \(N\)-Laplacian type with critical exponential growth in \(\mathbb{R} ^N\). J. Funct. Anal. 262(3), 1132–1165 (2012)
Li, X., Yang, M., Zhou, X.: Qualitative properties and classification of solutions to elliptic equations with Stein–Weiss type convolution part. Sci. China Math. 65, 2123–2150 (2022)
Li, Y., Ruf, B.: A sharp Trudinger–Moser type inequality for unbounded domains in \(\mathbb{R} ^{N}\). Indiana Univ. Math. J. 57, 451–480 (2008)
Lieb, E.H.: Existence and uniqueness of the minimizing solution of Choquard’s nonlinear equation. Stud. Appl. Math. 57, 93–105 (1977)
Lieb, E.H., Simon, B.: The Hartree–Fock theory for Coulomb systems. Commun. Math. Phys. 53, 185–194 (1977)
Lieb, E.H., Loss, M.: Analysis. In: Graduate Studies in Mathematics, AMS, Providence, Rhode Island (2001)
Lions, P.L.: The Choquard equation and related questions. Nonlinear Anal. 4, 1063–1073 (1980)
Lions, P.L.: The concentration-compactness principle in the calculus of variations. The limit case. I. Rev. Mat. Iberoam 1, 145–201 (1985)
Liu, X., Liu, J., Wang, Z.-Q.: Quasilinear elliptic equationss via perturbation method. Proc. Am. Math. Soc. 141, 253–263 (2013)
Liu, J., Wang, Y., Wang, Z.-Q.: Soliton solutions for quasilinear Schrödinger equations. II. J. Differ. Equ. 187(2), 473–493 (2003)
Liu, J., Wang, Y., Wang, Z.-Q.: Solutions for quasilinear Schrodinger equations via the Nehari method. Commun. Partial Differ. Equ. 29, 879–901 (2004)
Ma, L., Zhao, L.: Classification of positive solitary solutions of the nonlinear Choquard equation. Arch. Ration. Mech. Anal. 195, 455–467 (2010)
Mawhin, J., Willem, M.: Critical Point Theory and Hamiltonian System. Springer, New York (1989)
Moser, J.: A sharp form of an inequality by N. Trudinger. Indiana Univ. Math. J. 20, 1077–1092 (1970/1971)
Moroz, I.M., Penrose, R., Tod, P.: Spherically-symmetric solutions of the Schrödinger–Newton equations. Class. Quantum Gravity 15, 2733–2742 (1998)
Moroz, V., Van Schaftingen, J.: Groundstates of nonlinear Choquard equations: existence, qualitative properties and decay asymptotics. J. Funct. Anal. 265, 153–184 (2013)
Moroz, V., Van Schaftingen, J.: A guide to the Choquard equation. J. Fixed Point Theory Appl. 19, 773–813 (2017)
Medeiros, E., Severo, U.: On a quasilinear nonhomogeneous elliptic equation with critical growth in \(\mathbb{R} ^n\). J. Differ. Equ. 246, 1363–1386 (2009)
Miyagaki, O.H., Soares, S.H.: Soliton solutions for quasilinear Schrödinger equations with critical growth. J. Differ. Equ. 248(4), 722–744 (2010)
Pekar, S.I.: Untersuchung über die Elektronentheorie der Kristalle. Akademie Verlag, Berlin (1954)
Pohozaev, S.I.: The Sobolev embedding in the case \(pl = n\). In: Proc. Tech. Sci. Conf. on Adv. Sci., Research Mathematics Section, Moscow, 1965, pp. 158–170 (1964–1965)
Poppenberg, M., Schmitt, K., Wang, Z.-Q.: On the existence of soliton solutions to quasilinear Schrödinger equations. Calc. Var. Partial Differ. Equ. 14, 329–344 (2002)
Porkolab, M., Goldman, M.V.: Upper-hybrid solitons and oscillating two-stream instabilities. Phys. Fluids. 19, 872–881 (1978)
Ruf, B.: A sharp Trudinger–Moser type inequality for unbounded domains in \(\mathbb{R} ^2\). J. Funct. Anal. 219, 340–367 (2005)
Ruiz, D., Siciliano, G.: Existence of ground states for a modified nonlinear Schrödinger equation. Nonlinearity 23(5), 1221–1233 (2010)
Severo, U., Gloss, E., da Silva, E.: On a class of quasilinear Schrödinger equations with superlinear or asymptotically linear terms. J. Differ. Equ. 263(6), 3550–3580 (2017)
Shen, L., Rădulescu, V.D., Yang, M.: Planar Schrödinger–Choquard equations with potentials vanishing at infinity: The critical case. J. Differ. Equ. 329, 206–254 (2022)
Silva, E.A., Vieira, G.F.: Quasilinear asymptotically periodic Schrödinger equations with critical growth. Calc. Var. Partial Differ. Equ. 39, 1–33 (2010)
Stein, E.M., Weiss, G.: Fractional integrals on \(n\)-dimensional Euclidean space. J. Math. Mech. 7, 503–514 (1958)
Trudinger, N.S.: On imbeddings into Orlicz spaces and some applications. J. Math. Mech. 17, 473–484 (1967)
Willem, M.: Minimax Theorems. Birkhäuser, Boston (1996)
Yang, M., Rădulescu, V.D., Zhou, X.: Critical Stein–Weiss elliptic systems: symmetry, regularity and asymptotic properties of solutions. Calc. Var. Partial Differ. Equ. 61, 109 (2022)
Yang, M., Zhou, X.: On a coupled Schrödinger system with Stein–Weiss type convolution part. J. Geom. Anal. 31, 10263–10303 (2021)
Zhang, C., Chen, L.: Concentration-compactness principle of singular Trudinger–Moser inequalities in \(\mathbb{R} ^n\) and \(n\)-Laplace equations. Adv. Nonlinear Stud. 18(3), 567–585 (2018)
Zhang, Y., Tang, X.: Large perturbations of a magnetic system with Stein-Weiss convolution nonlinearity. J. Geom. Anal., 32(3), Paper No. 102, 27 pp (2022)
Zhang, Y., Tang, X., Rădulescu, V.D.: Anisotropic Choquard problems with Stein-Weiss potential: nonlinear patterns and stationary waves. C. R. Math. Acad. Sci. Paris 359, 959–968 (2021)
Author information
Authors and Affiliations
Corresponding authors
Ethics declarations
Conflict of interest
The authors declare that there is no conflict of interest. We also declare that this manuscript has no associated data.
Additional information
Communicated by Gerald Teschl.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
C. O. Alves is partially supported by CNPq/Brazil 307045/2021-8 and Projeto Universal FAPESQ-PB 3031/2021.
L. J. Shen is partially supported by NSFC (12201565).
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Alves, C.O., Shen, L. Soliton solutions for a class of critical Schrödinger equations with Stein–Weiss convolution parts in \(\mathbb {R}^2\). Monatsh Math 205, 1–54 (2024). https://doi.org/10.1007/s00605-024-01980-0
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00605-024-01980-0
Keywords
- Quasilinear Schrödinger equations
- Stein–Weiss convolution
- Trudinger–Moser inequality
- Change of variable
- Nash–Moser iteration
- Asymptotical behavior